From ddd3393b255c8d36f7901303994459f83fb8766a Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 7 Sep 2026 15:53:13 +0000 Subject: [PATCH 1/4] =?UTF-8?q?board:=20plan=20inventory=202026-09-07=20?= =?UTF-8?q?=E2=80=94=20211=20plans,=20V3=20waves,=20W-waves/jc=20vs=20code?= =?UTF-8?q?;=20V3=20folder=20now=20sees=20the=20harvest?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit - .claude/board/PLAN-INVENTORY-2026-09-07.md (new): 149 OPEN / 13 CLOSED / 9 SUPERSEDED / 40 AMBIGUOUS; V3 W0–W6 verdicts; ENTROPY M1–M27; W1a 5/5 shipped, W1b 0/6 closed (179 raw-intrinsic lines), W1.5 #6/#7 shipped + consumed, #8 ungated; jc registry facts (no Pillar 6; Pillar 11 feature- gated); top-level .claude census (71 files, 9 orphans); name collisions; synergy digest; recommended next order. Mints no D-ids (a snapshot). - evidence: five verbatim read-only agent tag-files under exec-runs/plan-inventory-2026-09-07-{plans-a-k,plans-l-z,v3-and-harvest, w-waves-and-jc,toplevel-and-board}.md; every claim entering a board file re-verified at a tree line (one agent claim corrected: the "two dangling INTEGRATION_PLANS paths" are cross-repo citations) - TECH_DEBT (append-only regrades): W1a #1–#5 → SHIPPED (primitive side; consumer halves stay Open); W1.5 #7 → SHIPPED + consumed (sigker/src/randomized.rs:49,124); TD-SIMD-SWEEP-W2 half done (types.rs clean, ndarray_bridge.rs:465,493 not); #8 ungated note - STATUS_BOARD: D-LNC-5a / D-MW-P2 → Shipped (#1198 merged 3797237b) - .claude/v3 (did not reference .claude/nexgen or the 2026-09-01 literature harvest): README doc-map row + two-collision note + rulings added to the canonical list; primer §5 ⊘ D-CCF-4 RESCINDED (on STATUS_BOARD since 2026-07-03, in no V3 file) + §6 no-exceptions row; INTEGRATION-PLAN W6 ⊘; routing.md §5 ⊘; compiled-templates StepMask struck; ENTROPY M2 → SHIPPED, M24 regrade, M27 added; COMPONENT-MAP §6 NestedBands row; FUTURE-DESIGN 2026-09-07 block (records the W2a Addendum-12a vs -15 contradiction, not adjudicated); witness-nibble-lane P5 (lit-harvest D3) - nexgen plan: reverse cross-link to .claude/v3 - EPIPHANIES ×4 (inventory finding; the operator's per-domain-table / chain-of-masks ruling; seven harvest sources → one version-keyed mask set; rung × band × plasticity are three axes), IDEAS ×5, LATEST_STATE delta, AGENT_LOG run record - counterfactual-rung3-closure-v1: FUTURE-DESIGN line-range citation → stable anchor (the prepend moved it; citation-decay gate) - gates: append_only_gate origin/main OK (9 files); citation_decay --since origin/main 0 new; supersession index regenerated last (no diff) Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp --- .claude/board/AGENT_LOG.md | 9 + .claude/board/EPIPHANIES.md | 60 ++ .claude/board/IDEAS.md | 40 ++ .claude/board/LATEST_STATE.md | 8 + .claude/board/PLAN-INVENTORY-2026-09-07.md | 311 +++++++++ .claude/board/STATUS_BOARD.md | 4 +- .claude/board/TECH_DEBT.md | 18 +- .../plan-inventory-2026-09-07-plans-a-k.md | 336 +++++++++ .../plan-inventory-2026-09-07-plans-l-z.md | 580 ++++++++++++++++ ...inventory-2026-09-07-toplevel-and-board.md | 648 ++++++++++++++++++ ...lan-inventory-2026-09-07-v3-and-harvest.md | 238 +++++++ ...lan-inventory-2026-09-07-w-waves-and-jc.md | 266 +++++++ .../nexgen-mask-histogram-thresholds-v1.md | 11 + .../plans/counterfactual-rung3-closure-v1.md | 5 +- .claude/v3/COMPONENT-MAP.md | 1 + .claude/v3/ENTROPY-MILESTONES.md | 5 +- .claude/v3/FUTURE-DESIGN.md | 10 + .claude/v3/INTEGRATION-PLAN.md | 9 + .claude/v3/README.md | 17 +- .claude/v3/knowledge/compiled-templates.md | 2 +- .claude/v3/knowledge/v3-substrate-primer.md | 7 + .claude/v3/soa_layout/routing.md | 8 +- .claude/v3/soa_layout/witness-nibble-lane.md | 9 + 23 files changed, 2584 insertions(+), 18 deletions(-) create mode 100644 .claude/board/PLAN-INVENTORY-2026-09-07.md create mode 100644 .claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md create mode 100644 .claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md create mode 100644 .claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md create mode 100644 .claude/board/exec-runs/plan-inventory-2026-09-07-v3-and-harvest.md create mode 100644 .claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md diff --git a/.claude/board/AGENT_LOG.md b/.claude/board/AGENT_LOG.md index 936b12bc8..305f334cf 100644 --- a/.claude/board/AGENT_LOG.md +++ b/.claude/board/AGENT_LOG.md @@ -1,3 +1,12 @@ +## 2026-09-07 — plan inventory: 5 read-only Sonnet agents, orchestrator-verified, consolidated to `PLAN-INVENTORY-2026-09-07.md` + +- **Why:** operator: *"create an inventory about the plans in `.claude` `.claude/V3` `.claude/plans` — use sonnet agent for grindwork — what is still open what is closed — check if v3 already has the harvest and update — write down any epiphanies and expansion ideas"*, plus *"check the synergies of mississippi queen board game blasgraph shannon EWA known unknowns ternlogq etc"*, plus the per-domain-table / chain-of-masks / SPOG-bakes dialogue. +- **Agents (all `general-purpose` Sonnet, read-only, no cargo, no GitHub writes, each writing ONLY its own tag-file — the orchestrator is the sole writer of every board file):** plans A–K (103 files) · plans L–Z (108 files, 27 special-attention reads in full) · `.claude/v3/` 21 files + harvest cross-check (scope widened mid-run to `.claude/nexgen/` + the 2026-09-01 literature harvest) · W1a/W1b/W1.5 + jc pillars vs code (ndarray `master` `b9afcb9b`) · top-level `.claude/*.md` (71) + STATUS_BOARD/ISSUES/TECH_DEBT/INTEGRATION_PLANS/LATEST_STATE/IDEAS/EPIPHANIES/SUPERSESSION-INDEX census. Tag-files banked verbatim: `exec-runs/plan-inventory-2026-09-07-{plans-a-k,plans-l-z,v3-and-harvest,w-waves-and-jc,toplevel-and-board}.md`. +- **Orchestrator verification (every claim that entered a board file):** confirmed — #1198 merged `3797237b` vs two "In PR" rows; the five W1a symbols at their cited lines; `sigker/src/randomized.rs:49,124` consumes the #7 primitive (the agent had hedged "not wired" — it IS); `types.rs:462` vs `ndarray_bridge.rs:465,493`; Pillar 11 default-deferred at `hambly_lyons.rs:737`; D-CCF-4 rescission at STATUS_BOARD:950; Addendum-15 at INTEGRATION-PLAN:744; `CLAUDE.md:762,765`; lit-harvest D3 wording at `:104`. **Reshaped one:** the "two dangling INTEGRATION_PLANS paths" are cross-repo citations (`:2018` tesseract-rs, `:2339` ndarray), not missing local plans. +- **Synthesis on the main thread (accumulation):** synergy read across `gemm-ternlog-mask-consolidation-v1` §9–§13, `membrane-tiers.md`, `SYNERGY-MAP-S00-S07`, `entropy-closure-causal-ground-v1`, `ew64-witness-unification-v1`, `epistemic-quadrant-materialization-v1`, `mul-ewa-trust-propagation-v1`, `rubicon-loco-rung-cognitive-fabric-v1`, `jc/src/ewa_sandwich.rs`, `spog_tenants.rs`, the four nexgen reader reports, the nexgen plan, and medcare-rs `obo_store.rs`/`domain_block.rs`/`orphanet.rs`/`rails.rs`/`state.rs` for the read-time re-derivation sites. +- **Board writes (this PR):** `PLAN-INVENTORY-2026-09-07.md` (new); TECH_DEBT 7 regrades (W1a ×5 SHIPPED-primitive-side, W1.5 #7 SHIPPED+consumed, TD-SIMD-SWEEP-W2 half-done, #8 ungated note); STATUS_BOARD D-LNC-5a/D-MW-P2 → Shipped; `.claude/v3/` 9 mirrors (README doc-map row + collision note + ruling list; primer §5 ⊘ + §6 row; INTEGRATION-PLAN ⊘; routing.md ⊘; compiled-templates StepMask struck; ENTROPY M2/M24 regrades + M27; COMPONENT-MAP §6 `NestedBands` row; FUTURE-DESIGN 2026-09-07 block; witness-nibble-lane P5); nexgen plan reverse link; EPIPHANIES ×4; IDEAS ×5; LATEST_STATE delta. Gates: `append_only_gate.py origin/main`, `citation_decay.py --since`, supersession index regenerated LAST. +- **Not done, on purpose:** no code, no plan-header edits in other owners' plans (recorded in §1a instead), no adjudication of the W2a contradiction, no force-push anywhere (the designated branch's remote tip was already-merged history, so the push is a fast-forward). + ## 2026-09-06 — temporal audit arc: 12 read-only agents + 2 probes, orchestrator-gated (#1198) - **Why:** operator asked for an archaeology + migration audit of the alpha channel, the rung ladder and the octopus/SPOG addressing, then for the delta-backed alpha layer, then for the findings to be written down as a durable reference. diff --git a/.claude/board/EPIPHANIES.md b/.claude/board/EPIPHANIES.md index 551a2f91f..fc67fe4a3 100644 --- a/.claude/board/EPIPHANIES.md +++ b/.claude/board/EPIPHANIES.md @@ -1,3 +1,63 @@ +## 2026-09-07 — E-A-PLAN-INVENTORY-FINDS-THE-BOARD-LAGS-THE-TREE-IN-BOTH-DIRECTIONS-1 — status cells decay at the rate of the tree, not of the file + +**Status:** FINDING, measured (five read-only Sonnet agents; every claim below re-verified by the orchestrator at a tree line, not a tag-file line). Full census: `.claude/board/PLAN-INVENTORY-2026-09-07.md`; evidence: `exec-runs/plan-inventory-2026-09-07-*.md`. +**Confidence:** High for every `file:line` row; the D-id coverage percentage is measured on the A–K half of `.claude/plans/` only. + +**The numbers.** 211 plans: 149 OPEN, 13 CLOSED, 9 SUPERSEDED, 40 AMBIGUOUS (35 with no status line at all — 22 of them the batch-landed `3DGS-*` design genre). V3 waves: W0 closed; W1 mechanism closed, adoption partial; W2 partial and self-contradictory; W3/W4 partial; W5/W6 open, W6's premise rescinded. W1a SIMD: 5/5 shipped. W1b: 0/6 closed (179 raw-intrinsic lines across five crates). W1.5: #6, #7 shipped and consumed, #8 unbuilt but ungated. + +**Direction one — docs still say Open for work on `main`.** Five `TECH_DEBT` rows said `Open` for primitives that have been on ndarray `master` for months (`simd_scalar.rs:1684,1709,1799,1887,1916`); W1.5 #7 said `Deferred` while `ndarray/src/hpc/randomized_signature.rs:292` ships it and `sigker/src/randomized.rs:124` consumes it — a THREE-way inconsistency, two knowledge docs having said SHIPPED since 2026-09-02; `STATUS_BOARD` D-LNC-5a/D-MW-P2 read `In PR #1198` with #1198 merged (`3797237b`); ENTROPY M2 said `QUEUED` nine weeks after D-PERT-1 shipped; `compiled-templates.md` said `StepMask` does not exist eight weeks after it did; `self-reasoning-substrate-v1.md:15` says "doc-only, no code" over four D-SRS rows that read Shipped with 147 tests between them. All but the last two are regraded in this PR (append-only strike-through with the tree line). + +**Direction two — docs still carry a rescinded premise or an unminted id.** `D-CCF-4` (the `0x1000` retirement) was RESCINDED on 2026-07-03 and sat on `STATUS_BOARD:950` for nine weeks while `INTEGRATION-PLAN.md:110`, `routing.md:104-106` and the primer §5 kept describing the marker as temporary with a 100 %-adoption trigger — no V3 file carried the rescission. W2a is stated two incompatible ways in ONE file (Addendum-12a: a new gated `BoardAggregates` tenant; Addendum-15: "SHIPPED as `ValueTenant::Kanban`", the per-ROW tenant its own sibling doc calls a sibling, not a substitute). `D-GRAPH-1`, `D-TRUTH-1`, `D-INFER-DEDUCTIONS-RELATION-BLIND` are cited by two plans and exist on no board row. + +**The gap the board already names, re-measured.** `ISS-PLAN-TRACKING-IS-UNENFORCED`: on the A–K half, **314 of 559 D-id citations (56 %) have no `STATUS_BOARD` row; 26 plans have zero coverage**. Two instrument traps, both caught: the regex `D-[A-Z]+(-[A-Z0-9]+)+` matches the tail of `E-READ-NOT-GREP` as `D-NOT-GREP` (81 false positives across 43 files); and the house style `D--N` is a *finding* citation, not a deliverable — "no board row" is evidence only after the id's grammar is checked. `INTEGRATION_PLANS.md` names 58 of 211 plan files nowhere. + +**The harvest was invisible from V3.** `.claude/v3/` referenced neither `.claude/nexgen/` nor the 2026-09-01 literature harvest; its every "nexgen" meant `openproject-nexgen-rs`. Nine mirrors landed (README row, primer §5/§6, INTEGRATION-PLAN, routing, compiled-templates, ENTROPY M27, COMPONENT-MAP §6, FUTURE-DESIGN, witness-lane P5) and the nexgen plan links back. + +**Verification reshaped one finding — which is why the agents' files are evidence, not verdicts.** The top-level agent reported "two cited plan paths that do not exist"; both are cross-repo citations (`INTEGRATION_PLANS.md:2018` → tesseract-rs, `:2339` → ndarray) misread as local. The agent that hedged "#7 not yet wired into sigker" was wrong in the OTHER direction — it is wired. The rule this leaves: **a status cell is a claim about the tree, so a regrade cites the tree (`file:line` on the branch), never the report that noticed it.** + +## 2026-09-07 — E-EVERY-DOMAIN-IS-A-TABLE-AND-A-CROSSWALK-IS-A-CHAIN-OF-MASKS-1 (OPERATOR-RULED) — no DataFusion joins, ever; the six blockers were read-time re-derivation + +**Status:** RULING (operator, 2026-09-07, verbatim below) + FINDING for the mechanism mapping (every site `file:line`) + CONJECTURE for the throughput claim until PROBE-CROSSWALK-MASK-1 runs (IDEAS 2026-09-07). +**Confidence:** High on the sites and on the bounds (all measured in `gemm-ternlog-mask-consolidation-v1` §12–§13 and D-GTM-0n/P3); medium on the unification until one probe joins two legs. + +**The ruling.** *"i mean every domain is just another table keyed by CUI STDID (snomed) loinc etc — its a chain effect with masking, no datafusion joins ever."* + +**What it dissolves.** Every open blocker in the SPOG / alpha / medcare / DataFusion arc is the same shape: **something decided once upstream is re-derived at read time, per row.** The ontology category is stamped into `value[96]` and read back per row (`medcare-cohorts/src/obo_store.rs:16-18,77,681`); `Domain::of_row` classifies per row (`orphanet.rs:165`, `rails.rs:432`); `FacetRegime::PerRowTui` resolves the regime per row (`domain_block.rs:259,695`, `quad_tenant.rs:169,215`); `graph_of(addr)` is right in direction (G read from the key, `contract/src/spog_tenants.rs:38`) but resolved per claim by a linear scan over tenants (`:87`); V1 external-id identity re-derives the row from an id the key already is; and Lance row identity is re-materialised by the scanner (`medcare-server/src/state.rs:900,936` `with_row_id = true`, `with_row_addr = true`). One table per G removes the first four reads (the domain IS the table); V3 lane-local addresses remove the fifth; containment removes the sixth. + +**The mechanism, named at T1.** The quad's 4×24 slots are four pre-resolved foreign keys — slot 0 the own key; CUI present in 8/8 domains as the hub; FMA the anatomy↔imaging bridge; ICD↔MONDO inside disease. A crosswalk is a chain: `eq_u32_to_mask` over table n's FK column (`ndarray/src/simd_int_ops.rs:562`), `mask_ternlog` with the incoming survivors (`:983`; `AND3` / `AND_ANDNOT2` by name, `membrane-tiers.md` T1), and the survivors' key set is the needle set of table n+1. **Nothing is joined: the mask-out of hop n IS the mask-in of hop n+1.** The one forbidden move is a mask-AND across two tables — that is a join in disguise, and it is exactly what the nexgen plan's room 18 already rejects at the seal ("a derived read that has to AND masks from two semantic families is a violation"). This is the schema-level form of the mask plan's own invariant (§11: *substrate == mask geometry == projection surface*). + +**The bounds that ride with the claim (none may be dropped when it is cited).** D-GTM-0j: masks win on Boolean relations at EVERY density (297× vs dense f32) — the boundary is TYPE, not density; a hop that must accumulate a value (evidence strength, NARS frequency) is not a mask hop and goes to the valued arm (blasgraph semirings over CSR; D-GTM-0g, the CSR SpMV arm, is unrun). D-GTM-0k: 0 bytes/step, by a counting allocator. D-GTM-0n/P3: chaining pays (`T3/T1` → 0.50 by K=8) only while the masks stay L2-resident (ratio back to 1.03 at a 512 KiB mask) and only above ~0.1 % active — a crosswalk hop that thins the survivors below that must switch to sparse. D-GTM-0l: packed-prefix routing does NOT carry long-range relations (1.08× at k=3 on physical addresses) — cross-domain hops ride the FK columns, never prefix locality between tables. `E-Q8`: six-neighbourness does no work — no neighbourhood structure between domains is assumed. + +**The unblock order (the dialogue's, confirmed against the tree).** (1) one table per domain — SPOG-aware bakes, one artifact per G; (2) V3 lane-local addresses (the mint side already bangs on V1: `E-A-V3-MINT-MUST-NEVER-DEGRADE-TO-V1-1`); (3) `ogar-r2il` gets its first consumer via `lance-graph-ogar` — today it has none (only `OGAR/Cargo.toml` names it); (4) DataFusion containment — both scanner flags off, pinned by a test, no new surface (`E-PLANNING-MIGRATES-TO-LOCO-R2IL-DATAFUSION-IS-GRACE-PERIOD-1`). Order matters: (4) before (1) leaves the consumer with no identity but the flags. + +**Falsifier.** PROBE-CROSSWALK-MASK-1: survivor SETS identical to the DataFusion path on one fixture; if they differ, the FK reading of the slots is wrong before the algebra is. + +## 2026-09-07 — E-SEVEN-HARVEST-SOURCES-ONE-OBJECT-THE-VERSION-KEYED-MASK-SET-1 — Mississippi Queen, Shannon, EWA, known unknowns, ternlogq and blasgraph are readings of `NestedBands` + +**Status:** SYNTHESIS [H] — each leg cites a measured or ruled source; the unification is unmeasured until a probe joins two legs (the first candidate: PROBE-CROSSWALK-MASK-1 + the `SettlementCell` classifier, IDEAS 2026-09-07). +**Confidence:** Medium. The legs are [G]/[H] individually; "one object" is the claim under test. + +- **Mississippi Queen** (`ndarray/.claude/plans/gemm-ternlog-mask-consolidation-v1.md` §9–§11): reveal-ahead-of-cursor (M1), tile-serves-every-boat with cache key `(mask generation, panel index)` (M1b), the ±1 lookahead ladder (M2), the coal budget for re-chaining (M3). The nexgen plan's room 10 says time travel is free because every `NestedBands` is version-keyed and never rewritten. **These are one cache: the mask generation IS the Lance version**, and the coal budget IS the rollover budget (D-NXG-5). Do not build a second invalidation scheme. The hexagon was regraded [S]→[H] and `E-Q8` shows degree-1 matches hex to four decimals at 5.5× less memory — what survives of the board is the reveal and the budget, not the tiling. +- **Shannon** (`entropy-closure-causal-ground-v1` §4; nexgen E-NXG-19): entropy measures WHERE closure is, never WHAT kind of hole (CE64 59–60) nor HOW it may be asserted (61–63); measured, entropy LAGS the popcount budget by five steps on a real shift and is the did-the-split-help read (0.834 → 0.912), not the timer. The literature harvest's greedy INFO_GAIN Ω(n/log n) bound adds: hop order by expected survivor reduction is near-optimal — a sort over popcounts, no planner. +- **EWA** (`jc/src/ewa_sandwich.rs`; `mul-ewa-trust-propagation-v1` §0a/§0b; lit-harvest #27 — three operations share the word): the sandwich is certified NUMERICALLY only (PSD ≥ 0.999, tightness 1.467×); "geometry decides WHERE, never WHAT" (F-MEP-0a) and the closed-graph monotonicity STOP (F-MEP-0d) are the fences. On a crosswalk hop the discrete Σ trace already exists: `popcount(survivors)`. EWA may rank which hop to explore; it may not elevate a survivor to a fact. +- **Known unknowns** (`entropy-closure` SettlementCells; `epistemic-quadrant-materialization-v1` §4c–§4e; `mul-ewa` "the pothole is a TYPED ABSENCE"): a known unknown is a hop whose survivor mask has popcount > 1 — alternatives = the mask, discriminators = the FK columns not yet swept, evidence needed = the sweep. Crystal / Glass / GroundedUnresolved / Fog read off `(popcount, discriminator present, FK grounded)`; ADMIT = the column exists, BUDGET = popcount, PERMIT = `ReasoningBand`, TEST = the sweep. The fork-return rule holds unchanged: only the elimination (the mask) comes back. +- **ternlogq** (`membrane-tiers.md`; OGAR #298; nexgen rooms 4/18/20): `TERNLOG 0x86` is the only surviving loco call and its value byte is the program; the eight named immediates are a complete mask ISA; room 18's family-separation rule is the mechanical "no joins". A crosswalk program = `RANK` + `TERNLOG` per hop. +- **blasgraph** (`SYNERGY-MAP-S00-S07` §8.0; D-GTM-0j): integer-only HDR semiring algebra, no splat; the boundary with masks is TYPE — Boolean relations are masks, valued relations are semirings over CSR (D-GTM-0g unrun decides where). K0..K7, the SPO 2³ triadic projections, are the eight Boolean masks of one quad row. + +**One object.** A sealed `NestedBands` + overlap matrix per `(classid, version)`, read by each source its own way — search, monitoring, rank, basin, drift, quorum, hop order, known-unknown census (nexgen room 26 said as much before the crosswalk reading existed). **The fence that keeps it one object and not a god-field:** rung, band, plasticity and "style as a byte string of immediates" are four different things (next entry). + +## 2026-09-07 — E-RUNG-BAND-AND-PLASTICITY-ARE-THREE-AXES-NEVER-ONE-LEVEL-FIELD-1 — level of processing, level of assertion, and how content changes do not share a byte + +**Status:** FENCE (operator caution 2026-09-07 — "rung 0–9 and bits 61..63 must never be folded into one level field" — on top of two prior rulings). +**Confidence:** High — three shipped carriers with three cardinalities. + +- **Rung 0–9 = level of PROCESSING**, a temporal horizon: the `RungLevel` discriminant (`cognitive_shader.rs:157`), consumed by `EpistemicMode::for_rung` (`planner/src/temporal.rs:87-97`: 0..=4 Strict, 5..=8 Aware, 9 Retro) — it decides which versions a reader may see. Ten values. +- **`ReasoningBand` bits 61..63 = level of ASSERTION**, Tarski permission (`entropy-closure-causal-ground-v1` §4; `causal-edge::layout`) — it decides what a settled closure may be asserted as. Eight values. `ISS-RUNG-VS-BAND-CARDINALITY-COLLISION` (2026-08-26) already records that the two share endpoint names with different cardinalities, and `ISS-REASONING-BAND-GATES-NOTHING` that the band gates no control loop yet — the temptation to "just use rung" is live. +- **`PlasticityState` 3-bit S/P/O = how content CHANGES** (`causal-edge/src/plasticity.rs`); `epistemic-quadrant-materialization-v1` §4b refused "STDP = rung 3" for exactly this reason — *a rung says what content IS, a plasticity mode says how it CHANGES; different axes* — and the substrate already agrees (per-plane hot/frozen is axis-local). +- The nexgen "thinking style = a byte string of immediates" (room 4) is a **fourth** thing, a program, not a level. + +**Consequence for the chain-of-masks reading:** PERMIT is the band; the readable horizon is the rung; the mask is neither. **Falsifier / gate:** any struct, enum or lane that stores two of the three in one field, or any `level: u8` that is read as more than one of them, is a LAYOUT-BREAK-class defect (`v3-envelope-auditor` verdict); a `/v3-audit` grep for such fields is the cheap pre-commit check (not yet added — recorded so the next session adds it rather than re-deriving the fence). + ## 2026-09-07 — E-PLUG-AND-PLAY-IS-THE-DECLARATION-NOT-A-TABLE-1 — my fix rebuilt the lockstep it was closing **Status:** FINDING, measured. Fixed in this PR (7 tests, both guards disable-verified). diff --git a/.claude/board/IDEAS.md b/.claude/board/IDEAS.md index 23098a6f5..185299cd3 100644 --- a/.claude/board/IDEAS.md +++ b/.claude/board/IDEAS.md @@ -87,6 +87,46 @@ Agents filter by `@`-mention or domain to see what's theirs. (Prepend new ideas here with today's date. Format:) +## 2026-09-07 — SPOG-aware bakes: one artifact per G, and the crosswalk as a chain of masked equality sweeps (PROBE-CROSSWALK-MASK-1) + +`[P1 @simd-savant @kernel-membrane-warden @layer-boundary-warden domain:medcare/mask D-NXG-2 D-NXG-12]` + +Operator (2026-09-07): *"every domain is just another table keyed by CUI STDID (snomed) loinc etc — its a chain effect with masking, no datafusion joins ever."* Mechanically: bake one artifact per G (`graph_of(addr)`, `spog_tenants.rs:38`) instead of stamping the category into `value[96]` and re-reading it per row (medcare `obo_store.rs:16-18,77,681`); read the quad's 4×24 slots as four pre-resolved foreign keys (slot 0 = own key; CUI present in 8/8 domains = the hub; FMA = the anatomy↔imaging bridge; ICD↔MONDO inside disease). A crosswalk is then `eq_u32_to_mask` (`ndarray/src/simd_int_ops.rs:562`) on the FK column of table n, `mask_ternlog` (`:983`) with the incoming survivor mask, and the survivors' key set becomes the needle set for table n+1 — never a mask-AND across two tables (nexgen room 18). **Probe, pre-registered:** on medcare's baked OBO tables, run CUI→SNOMED→LOINC as a mask chain and as the existing DataFusion path; assert (a) survivor SETS identical, (b) 0 bytes/step under the counting allocator (D-GTM-0k's instrument), (c) per-hop ns flat in chain length while every mask fits L2 (the D-GTM-0n bound rides with the claim), (d) a deliberately cross-family AND is REJECTED at the seal (can-it-fire), (e) a hop with < 0.1 % survivors is routed to the sparse arm (0n's other bound). Falsifier: (a) fails ⇒ the FK reading of the slots is wrong, not the mask algebra. + +**Status:** Open — needs the medcare-side bake shape first (step 1 of `PLAN-INVENTORY-2026-09-07.md` §9). + +## 2026-09-07 — DataFusion containment: pin `with_row_id`/`with_row_addr` OFF with a test that fails when either flips; no new surface + +`[P1 @layer-boundary-warden domain:infra]` + +`E-PLANNING-MIGRATES-TO-LOCO-R2IL-DATAFUSION-IS-GRACE-PERIOD-1` freezes DataFusion; it does not yet fence the one seam that re-derives row identity at read time: medcare `state.rs:900,936` passes `with_row_id = true` / `with_row_addr = true`, i.e. the scanner re-materialises what the 16-byte key already is. Containment = both flags off, one test that asserts the scan's schema carries neither `_rowid` nor `_rowaddr` (goes red if a future edit re-enables them), and a grep fence in `/v3-audit` for new `SessionContext`/`datafusion::` imports outside the grace-period files. Order matters: this is step 4, AFTER the per-G bakes exist, or the flags are the only identity the consumer has. + +**Status:** Open. + +## 2026-09-07 — `ogar-r2il` gets its first consumer through `lance-graph-ogar`: `RANK` + `TERNLOG 0x86` are the two calls a crosswalk program needs + +`[P2 @codegen-flow-cartographer @layer-boundary-warden domain:r2il D-R2IL-5 D-RLR-4]` + +`ogar-r2il` has zero consumers in lance-graph (only `OGAR/Cargo.toml` and its own manifest name it, 2026-09-07); `ogar_loco::TERNLOG = FnIndex(0x86)` is minted and unconsumed; the nexgen plan's room 20 wants a `RANK` macro (`where rank(col) <= n` → one AND against a cached band mask). A crosswalk program is exactly those two calls in sequence: `RANK` to admit, `TERNLOG` (immediate = the hop conjunction) per hop. Wire it via `lance-graph-ogar` (the armed tier, workspace-excluded, so the OGAR path dep cannot break workspace load). Falsifier: a lifted R2IL crosswalk program yields the same survivor mask bit-for-bit as the hand-written chain in PROBE-CROSSWALK-MASK-1. Attaches to D-R2IL-5 (the lgj membrane) — not a new owner, per the D-RLR-4 HELD ruling. + +**Status:** Open (blocked on PROBE-CROSSWALK-MASK-1's chain existing). + +## 2026-09-07 — Hop order by popcount: the greedy INFO_GAIN bound turns the crosswalk planner into a sort + +`[P2 @truth-architect domain:mask D-NXG-12]` + +The literature harvest banks a greedy INFO_GAIN Ω(n/log n) bound; the nexgen plan's room 23 stores a B×B overlap matrix beside each histogram. Together: the order in which a crosswalk's FK sweeps run is the descending-selectivity order of their popcounts, corrected by the overlap matrix for correlated predicates — a sort over ≤ a dozen numbers, not a join planner. Measure against the DataFusion optimizer's join order on the same fixture (it should tie or beat it on hop count; if it loses, the overlap correction is the missing term, not a planner). + +**Status:** Open (after PROBE-CROSSWALK-MASK-1). + +## 2026-09-07 — A known unknown is a survivor mask with popcount > 1: `SettlementCell` from (popcount, discriminator column, grounded FK) — no new type + +`[P2 @truth-architect @dto-soa-savant domain:epistemics D-ECG-2 D-KUH-1]` + +`entropy-closure-causal-ground-v1`'s four `SettlementCell`s and its ADMIT/BUDGET/PERMIT/TEST walker have a mask-native reading on a crosswalk hop: popcount == 1 with the FK grounded = Crystal; popcount == 1 by inference = Glass; popcount > 1 with a discriminator column present but unmeasured = GroundedUnresolved (the *typed absence* of `mul-ewa-trust-propagation-v1` §0a); popcount > 1 with no discriminator = Fog. ADMIT = the FK column exists in the quad; BUDGET = the popcount budget (E-NXG-19: it leads entropy); PERMIT = `ReasoningBand` (bits 61..63); TEST = the equality sweep. Nothing new is minted — the classification is a read of `(popcount, has_discriminator, grounded_fk)`. Falsifier: on a fixture with one hop of each kind, the classifier must land each in its cell and stay silent on Crystal (the can-it-stay-silent twin). + +**Status:** Open. + ## 2026-09-05 — Σ-propagation: the batched `F64x8` kernel is the real lever; AMX/MKL was the wrong shape for a 2×2 `[P2 @savant-architect @sentinel-qa domain:codec D-OIF-4 D-OIF-5]` diff --git a/.claude/board/LATEST_STATE.md b/.claude/board/LATEST_STATE.md index 564825868..1159152bd 100644 --- a/.claude/board/LATEST_STATE.md +++ b/.claude/board/LATEST_STATE.md @@ -1,3 +1,11 @@ +## 2026-09-07 — plan inventory landed: `PLAN-INVENTORY-2026-09-07.md` + the V3 folder now sees the harvest + +- **New board document:** `.claude/board/PLAN-INVENTORY-2026-09-07.md` — 211 plans (149 OPEN / 13 CLOSED / 9 SUPERSEDED / 40 AMBIGUOUS), V3 waves W0–W6, ENTROPY M1–M27, W1a/W1b/W1.5 + jc pillars vs code, top-level `.claude/*.md` (71; 9 orphans), board dashboards. Mints no D-ids (a snapshot, not a plan). Evidence: five verbatim tag-files under `exec-runs/plan-inventory-2026-09-07-*.md`. +- **Status cells corrected (append-only, strike-through + date):** TECH_DEBT W1a #1–#5 → SHIPPED (primitive side; consumer halves stay Open); W1.5 #7 → SHIPPED + consumed (`sigker/src/randomized.rs:124`); TD-SIMD-SWEEP-W2 → half done (`types.rs` clean, `ndarray_bridge.rs:465,493` not); STATUS_BOARD D-LNC-5a / D-MW-P2 → Shipped (#1198 `3797237b`); `.claude/v3/ENTROPY-MILESTONES.md` M2 → SHIPPED, M24 regrade, **M27 added** (five threshold vocabularies → `NestedBands`). +- **`.claude/v3/` now references the harvest** (it referenced neither `.claude/nexgen/` nor the 2026-09-01 literature harvest; "nexgen" inside it meant `openproject-nexgen-rs`): README doc-map row + two-collision note + `E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` / `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1` added to the ruling list; primer §5 ⊘ (D-CCF-4 RESCINDED — it had been on STATUS_BOARD since 2026-07-03 and in no V3 file) + §6 no-exceptions row; INTEGRATION-PLAN W6 ⊘; routing.md §5 ⊘; compiled-templates `StepMask` struck; COMPONENT-MAP §6 `NestedBands` row; FUTURE-DESIGN 2026-09-07 block (records the W2a Addendum-12a-vs-15 contradiction without adjudicating it); witness-nibble-lane P5 (lit-harvest D3 cross-ref). The nexgen plan links back. +- **Contract inventory — net delta:** none. No type, module or lane changed; `NestedBands` (`planner/src/nested_bands.rs`, D-NXG-1, 2026-09-05) is now *mapped* in COMPONENT-MAP §6, not new. +- **Epiphanies:** four 2026-09-07 entries (inventory finding; the operator's per-domain-table / chain-of-masks ruling; the seven-source mask-set synthesis; rung × band × plasticity are three axes). **Ideas:** five 2026-09-07 entries (SPOG-aware bakes + crosswalk probe; DataFusion containment pin; `ogar-r2il` first consumer via `RANK` + `TERNLOG`; greedy hop order by popcount; known-unknown = survivor mask with popcount > 1). + ## 2026-09-07 — D-BLOCKS-HOTPLUG-1 (operator ruling): a consumer is NOT a canon builtin — and OGAR had already ruled it in code **Ruling, operator, 2026-09-07.** *"Blockly is a hot-plugged consumer, not a diff --git a/.claude/board/PLAN-INVENTORY-2026-09-07.md b/.claude/board/PLAN-INVENTORY-2026-09-07.md new file mode 100644 index 000000000..c60fb4a37 --- /dev/null +++ b/.claude/board/PLAN-INVENTORY-2026-09-07.md @@ -0,0 +1,311 @@ +# Plan inventory — 2026-09-07 (`.claude/`, `.claude/v3/`, `.claude/plans/`, W-waves, jc) + +> **What this is.** A measured census of every plan file, the V3 folder, the +> SIMD W-waves and the jc pillar registry, against the tree at `aeebfb23` +> (main, 2026-09-07). Five read-only Sonnet agents wrote the evidence +> (verbatim tag-files under `exec-runs/plan-inventory-2026-09-07-*.md`); the +> orchestrator spot-checked every claim that enters a board file below +> (`file:line` in the tree, not in the tag-file) and corrected one. This is +> a snapshot, not a plan: it mints **no D-ids**, so `supersession_index.py` +> and `plan_dids.py` do not see it — by design. +> +> **Operator ask (2026-09-07):** *"create an inventory about the plans in +> `.claude` `.claude/V3` `.claude/plans` … what is still open what is closed, +> check if v3 already has the harvest and update, write down any epiphanies +> and expansion ideas."* Epiphanies: `EPIPHANIES.md` 2026-09-07 block. Ideas: +> `IDEAS.md` 2026-09-07 block. V3 harvest mirrors: §4. +> +> **How to read a verdict.** OPEN = the plan's own status line and/or its +> D-ids on `STATUS_BOARD.md` say work remains. CLOSED = the plan's own +> deliverable is delivered (a measurement, an audit, a shipped ladder). +> SUPERSEDED = a higher-numbered sibling exists or the status line says so. +> AMBIGUOUS = no status line in house format and no board row to decide it. +> Every verdict came from a **read of the status line in context** — naive +> substring matching on "SHIPPED" produced ≥ 6 false positives in this corpus +> (`genetic-research-substrate-integration-v1.md` "No code shipped yet", +> `weather-soa-bake-v1.md` "No code shipped", `entropy-ladder-spo-rung-v1.md` +> "R1 SHIPPED; R2–R6 planned", `mul-consumer-build-gate-v1.md` "GATE RUN" with +> §4 "OPEN — not discharged", …), exactly the trap `CLAUDE.md` § "What the +> gate does NOT prove" records for the `ARCHIVE?` route. + +--- + +## 0. The numbers + +| Surface | Count | Open | Closed | Superseded | Ambiguous | +|---|---|---|---|---|---| +| `.claude/plans/*.md` A–K (entries 1–103) | 103 | 64 | 1 | 2 | 36 | +| `.claude/plans/*.md` L–Z (entries 104–211) | 108 | 85 | 12 | 7 | 4 | +| **`.claude/plans/` total** | **211** | **149** | **13** | **9** | **40** | +| `.claude/v3/` waves W0–W6 | 7 | 4 (W3/W4 partial, W5, W6) | 1 (W0) | — | 2 (W1 mechanism-closed/adoption-partial; W2 contradictory) | +| `.claude/v3/ENTROPY-MILESTONES.md` M1–M27 | 27 | 9 queued + 6 in-flight + 1 ruling-needed | ~8 | — | — | +| `.claude/*.md` top-level | 71 | — | — | 1 (`SESSION_FALKORDB_CROSSCHECK`) | 9 orphans, 61 pre-June-2026 | +| `STATUS_BOARD.md` D-id rows | 592 (89 sections) | 77 Queued, 23 Blocked, 5 In PR | 33 Shipped | 1 | 427 narrative-status cells | +| `ISSUES.md` | 47 | 37 | 8 (+1 resolving) | 1 | — | +| `TECH_DEBT.md` | 167 | 95 | 7 (4 Paid + 3 Resolved) | — | 65 unlabeled (pre-Kanban-format era) | +| W1a SIMD primitives (ndarray) | 5 | 0 | 5 | — | — | +| W1b consumer migrations | 6 | 6 (one half-done) | 0 | — | — | +| W1.5 sigker primitives | 3 | 1 (#8, ungated) | 2 (#6, #7) | — | — | +| jc pillar registry | 12 | 1 deferred (Pillar 2) + 1 feature-gated (Pillar 11) | 10 execute by default | — | — | + +The 40 AMBIGUOUS plans are 35 A–K files with no `Status:` line at all (22 of +them the `3DGS-*` / `Palette256-3DSB-*` / `PhiSpiral256-SoA-*` design-doc +genre, landed in one batch on 2026-05-25; 13 ordinary plans that never got a +header), `cascade-seal-register-grid-v1.md` ("RATIFIED v3 — council complete" +with both D-ids absent from the board), and four L–Z files +(`ogit-g-context-bundle-v1`, `probe-revision-attention-view-1`, +`supabase-subscriber-v1`, `tarski-markov-hhtl-seam-v1` — the last is "NOT a +plan" by its own header). + +## 1. Board lags the tree — in BOTH directions + +The load-bearing finding. Fixed in this PR where the fix is a status cell; +recorded where it needs a reader. + +### 1a. Docs still saying *Open / Deferred / Queued / does not exist* for work that is on `main` + +| Where | Said | Tree | Fixed here | +|---|---|---|---| +| `TECH_DEBT.md` TD-NDARRAY-SIMD-{UNPACK-I4-16D, SATURATING-ABS-I8, GATHER, PREFETCH, POPCOUNT-U64} (W1a #1–#5) | `Status: Open` ×5 | every symbol on ndarray `master` `b9afcb9b` — `simd_scalar.rs:1684,1709,1799,1887,1916` / `simd_avx512.rs:2673,2716,2846,3091,2987` / `simd_neon.rs:2269,2301,2356,2445` | yes — regraded SHIPPED (primitive side); consumer halves stay Open | +| `TECH_DEBT.md` TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (W1.5 #7) | `DEFERRED` (heading + body) | `ndarray/src/hpc/randomized_signature.rs:96,205,244,292` (PR #294) **and consumed** — the import `use ndarray::hpc::randomized_signature::randomized_signature_sweep` at `crates/sigker/src/randomized.rs:49`; call site `RandomizedSignatureBuilder::encode` (`randomized.rs:124`) | yes — a three-way inconsistency (two knowledge docs said SHIPPED since 2026-09-02) | +| `TECH_DEBT.md` TD-SIMD-SWEEP-W2 | "`types.rs` (22 raw ops)" | `types.rs` has **0** raw-intrinsic lines (`hamming_distance_dispatch` → `ndarray::hpc::bitwise::hamming_distance_raw`, `types.rs:462`); `ndarray_bridge.rs` still 66, incl. `_mm512_popcnt_epi64` at `:465,:493` | yes — HALF DONE | +| `STATUS_BOARD.md` D-LNC-5a, D-MW-P2 | `**In PR** #1198` | #1198 merged `3797237b` | yes — Shipped | +| `.claude/v3/ENTROPY-MILESTONES.md` M2 | `QUEUED (small, mechanical)` | STATUS_BOARD D-PERT-1 Shipped #630, verified in-code 2026-07-10 — nine weeks before the file was last edited | yes | +| `.claude/v3/ENTROPY-MILESTONES.md` M24 | `PARTIAL 2026-07-17` | STATUS_BOARD D-MBX-A6-P3d/P4 (2026-08-01/02) ship the durable witness + recovery + cycle/WAL closure with the 64k/17 falsifier | yes — regrade appended | +| `.claude/v3/knowledge/compiled-templates.md` gap list | "`StepMask` type does not exist yet" | `contract::step_mask::StepMask`, D-V3-W3a, 2026-07-10 | yes | +| `.claude/plans/self-reasoning-substrate-v1.md:15` | "PROPOSED — doc-only. No code" | D-SRS-1..4 all **Shipped** on STATUS_BOARD (`reason.rs` + 7 tests; `shape,ancestry.rs` + 63; falsifier fired; `introspect.rs` + 77) | no — plan header, left for the owner; recorded | +| `LATEST_STATE.md` static sections `## Current Contract Inventory (lance-graph-contract)` (`:2915`), `## Immediate Next Work` (`:3073`), `## Deferred (…)` (`:3102`) | frozen at 2026-05-16 / 2026-06-26; cite D-CSV-13b "IN PR" (shipped long since) and keep `TD-COLLAPSE-GATE-SMALLVEC-1` in "Deferred" while self-annotating it "CLOSED 2026-06-11" (`:3106`) | ~100 dated deltas prepended above them since | no — the file is append-only; a reader must treat the top-of-file deltas as current, the static sections as a 2026-06 snapshot | +| `CLAUDE.md:762,765` | "`.claude/*.md` (61 top-level docs) … `SESSION_CAPSTONE.md`" | 71 files; no `SESSION_CAPSTONE.md` exists | no — CLAUDE.md is operator-owned; recorded | + +### 1b. Docs still carrying a premise that was rescinded, or ids that were never minted + +| Where | Said | Truth | Fixed here | +|---|---|---|---| +| `.claude/v3/INTEGRATION-PLAN.md:110` (D-CCF-4 row), `soa_layout/routing.md:104-106`, `knowledge/v3-substrate-primer.md:103-105` | `0x1000` is temporary; adoption 100 % ⇒ P4 ⇒ marker retires | **D-CCF-4 RESCINDED** (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) — STATUS_BOARD row 950 carried it for nine weeks; none of the three V3 files did | yes — ⊘ notes at all three | +| `.claude/v3/INTEGRATION-PLAN.md` Addendum-12a (`:486-510`) vs Addendum-15 (`:744`) | W2a = a NEW gated `BoardAggregates` tenant **vs** W2a "SHIPPED as `ValueTenant::Kanban`" | `ValueTenant::Kanban` is the pre-existing per-ROW tenant #9 that `mailbox-kanban-model.md` calls a *sibling, not a substitute*; `tenants.md` (2026-08-23) lists no `BoardAggregates` | no — recorded on FUTURE-DESIGN 2026-09-07; needs `canonical_node.rs:1622+` + owner wiring read, not a doc pass | +| `.claude/v3/COMPONENT-MAP.md` §4 W2b row | supervisor `KanbanActor` EXTEND | actor half DELETED 2026-08-05 (⊘ note already in the row since 2026-09-05); the wave table's W2b row still reads as an open work item | no — ⊘ exists; the wave table row is append-only history | +| `persistent-nars-kg-v1.md`, `self-reasoning-substrate-v1.md` | cite `D-GRAPH-1`, `D-TRUTH-1`, `D-INFER-DEDUCTIONS-RELATION-BLIND` | **0** hits on STATUS_BOARD — never minted or renamed without a pointer | no — recorded | +| `weather-substrate-poc-v2.md` | `D-WXA-*`/`D-WXB-*`/`D-WXC-*` ladder | never on STATUS_BOARD (`grep -c WXA` = 0); `weather-soa-bake-v1.md` §0.1 records it from inside the corpus | no — already recorded there | + +### 1c. The gap the board already names, re-measured + +`ISS-PLAN-TRACKING-IS-UNENFORCED` (`ISSUES.md:183`, 2026-09-03, OPEN) says +plans can mint D-ids the board never tracks. Measured on the A–K half: +**314 of 559 D-id citations (56 %) across 61 plans have no STATUS_BOARD +row; 26 plans have zero board coverage** (full list in +`exec-runs/plan-inventory-2026-09-07-plans-a-k.md` § Mismatches). The L–Z +half found the same shape but also the house style that makes a raw count +overstate it: `D--N` is a legitimate in-plan *finding* +citation, not a deliverable row, and the naive regex `D-[A-Z]+(-[A-Z0-9]+)+` +also matches the tail of `E-READ-NOT-GREP` as `D-NOT-GREP` (81 such false +positives across 43 files, filtered). So: "no board row" is evidence of a +tracking gap only after checking whether the id is grammatically a +deliverable. `plan-dids.yml` gates ADDED plans only; the backlog is a +cross-session scope call the workflow itself defers. + +`INTEGRATION_PLANS.md` cites 141 distinct `.claude/plans/` paths; **58 of +211 plan files are named nowhere in it** — 20 `3DGS-*` (own sub-index +`3DGS-PLAN-INDEX.md`), 5 `tesseract-rs-*`, 3 `alpha-reason-witness-shader- +field-*`, and 30 singles (list in `exec-runs/…-toplevel-and-board.md`). The +"two dangling cited paths" the agent reported are **not dangling** — they are +cross-repo citations (`tesseract-rs/.claude/plans/pdf-to-text-ocr-v1.md`, +`INTEGRATION_PLANS.md:2018`; `ndarray/.claude/plans/splat-native-ultrasound- +simd-substrate-v1.md`, `:2339`) misread as local. Verification reshaped the +finding; the tag-file keeps the original claim. + +## 2. `.claude/plans/` — what is genuinely open, by family + +Only plans whose OPEN state is load-bearing for a next session are named; +the two tag-files carry every row. + +| Family | Open on the board | Closed / shipped | The one thing a session should know | +|---|---|---|---| +| **V3 substrate** (`v3-substrate-integration-v1` stub → `.claude/v3/`, `bindspace-mailbox-soa-wiring-v1`, `bindspace-singleton-to-mailbox-soa-v1`, `cycle-coherent-soa-snapshot-v1`, `singleton-to-snapshot-nudge-v1`) | D-BSW-0..4 Queued/Blocked; D-SOA-SNAP-1..6 no board rows; D-SNGL-3 "In progress" since 2026-06-13 | shim built (`BackingStoreWrite`, 9 methods, zero callers) | `BindSpace` is RETIRE with **68 crate files / 47 plans / 41 blind** (`SUPERSESSION-INDEX.md`); retirement is D-BSW-4, deliberately BLOCKED — never a worker task | +| **Alpha / rung / witness** (`alpha-channel-rung-overlay-v1` 26 D-ids, 12 on board; `alpha-reason-witness-*` ×4; `rubicon-loco-rung-cognitive-fabric-v1`; `known-unknown-handover-network-v1`; `entropy-closure-causal-ground-v1`; `ew64-witness-unification-v1` 2/18 on board) | D-ACR-2 rail UNMINTED; D-ACR-3 no ontology write path (`mailbox_owner()` zero external callers — survives #1112); D-RLR-0..6 all Queued/HELD; D-ECG-*, D-EWU-2..9 Queued | D-ARW-0 Shipped (cited by 4 plans that all say "PLAN ONLY" — the work landed outside them); D-ACR-1/7/8 Shipped; `alpha`/`alpha_tunnel`/`rung_schedule` in contract (#1112) | the temporal audit (`.claude/temporal/`, #1198) FALSIFIED the migration hypothesis: `SpogTenants`, `AlphaTunnel`, `rung_horizon`, `AlphaMask`, `residue_band` are built, tested and **connected to nothing** — the open item is a first consumer, not a mechanism | +| **MUL / EWA / trust** (`mul-calibration-not-verdict-v1`, `mul-ewa-trust-propagation-v1`, `mul-consumer-build-gate-v1`) | D-MCAL-0..5, D-GATE-1..5, D-MEP-0/1 Queued; F-MUL-6 second half "OPEN — not discharged" (MedCare-rs never built against the head, `mul-consumer-build-gate-v1.md:174`) | `mul-consumer-census-v1` CLOSED; D-MCAL-1/6 Shipped | zero MUL↔EWA wiring exists; F-MEP-0b (sign of Σ) and F-MEP-0d (closed-graph monotonicity) are the STOP gates before any run; `ISS-KANBAN-PLAN-EXIT-HAS-NO-NAMED-ROUTE` (`Plan = 4` reachable by no primitive) is the operator's call | +| **Plasticity / Sudoku** (`epistemic-quadrant-materialization-v1`) | P1–P4 falsifiers unrun; `PlasticityState` inert (11 ALL_HOT sites, nothing consumes the bit); `TD-FORK-CANNOT-CLOSE-WHAT-SINGLES-CANNOT` | G1–G7 GREEN on engineered fixtures (a mechanism demonstrator, not a solver) | `PlasticityEngine` verdict REIMAGINE — tenant 7 already IS the Hebbian counter; §4b refuses "rung = plasticity mode" (a matrix, not a mapping) — see the rung/band/plasticity entry | +| **Signatures / sigker / jc** (`pillar11-signature-certification-unification-v1`, `jc-pillars-runtime-wiring-v1`) | W5 PowerSig DEFERRED by its own trigger (4609 in-tree vs ~11 585 memory-half, ~33 388 time-half) | W0–W4 SHIPPED; pillar11 plan CLOSED | two "Pillar 11"s (jc `hambly_lyons` uniqueness vs ndarray `hpc::pillar::signature` stability) + `jc::solver_order` as a 13th battery outside the registry; the SIMD W1a/W1b/W1.5 and this plan's W0–W5 are DISJOINT numberings | +| **Mask / nexgen** (`.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1`, `mask-algebra-revision-read-v1`) | D-NXG-2/6/7/8/9/10/12 Queued, D-NXG-11 Blocked (σ_step placeholder); D-MAR-2 blocked on an operator ruling (§5) | D-NXG-1/4/5 shipped (`planner/src/nested_bands.rs`, E-NXG-21); D-MAR-1 Shipped; HIST-1/ROLL-1/FLOOR-1 green with three restatements | budget LEADS entropy by 5 steps (E-NXG-19); `mu+3σ` floor unreachable on a real column (E-NXG-20); rooms 4–27 PROPOSAL; the plan lives outside `supersession_index.py`'s scan path on purpose | +| **R2IL / loco** (`r2il-machine-semantic-contract-v1` 1427 lines, `r2il-bpe-typed-genetic-recombination-v1`, `probe-r2il-live-regfile-v1`) | Q2 unresolved, Q3 an external OGAR-mint gate, Q4 may proceed, Q5 data half in another repo; the 4 recombination operators + contract pass + v3 admission loop unbuilt (`revise_if_minority_wins` is `todo!()`, blocked on D-PERSONA-5) | live-regfile 18/18 GREEN locally; §7 falsifiers ran | the r2sleigh CI workflow has **never executed** (`total_count: 0`, unregistered runner) — every green number is local; Q8 falsified the hex-topology hypothesis (E-Q8: degree-1 ablation matches to 4 decimals at 5.5× less memory); `ogar-r2il` has **zero consumers** in lance-graph (only `OGAR/Cargo.toml` + its own manifest name it) | +| **Persistence / lance** (`lance-convergence-staged-migration-v1`, `persistence-cycle-wal-bootstrap-v1`, `persistence-artifact-backed-commit-v1`) | D-LNC-4/5/6 Queued; D-MBX-A6 upgrade Queued; only Phase A implemented | D-LNC-0..3 Shipped (lance 11 / lancedb 0.38); D-LNC-5a/D-MW-P2 Shipped (#1198) | `LanceCycleWriter::open` has zero callers repo-wide; native lance delta is an optional accelerator, snapshot comparison is the oracle (D-TEMPORAL-2) | +| **Consumers** (`lance-graph-in-{medcare,smb-office,woa}-rs-v1`, `unified-bridge-consumer-migration-v1`, `ogar-sink-in-and-consumer-bridge-removal-v1`, `super-domain-rbac-tenancy-v1` 40 D-ids) | Drafts from 2026-05-25 never advanced; D-SINK-2/3/5 unconfirmed; D-V3-W5a..i all Queued | the six `*Bridge`s are `#[deprecated]` aliases over `UnifiedBridge

` — that half is DONE | smb-office-rs `LanceConnector::upsert` is still the ONE live orphan online write (W5f); `ClassRbac × ClassView × WideFieldMask` is unenforced across nine repos (D-OIF-1-DEC withdrawal) | +| **Odoo / OGIT / ontology** (`odoo-*` ×6, `lance-graph-ontology-v5`, `normalized-entity-holy-grail-v1`, `ogit-*`) | D-ODOO-SAV-5a..e Queued; 14/15 D-ONTO-V5 unconfirmed; D-CASCADE-V1-1..15 "plan, not implementation" | `odoo-source-extraction-v1` CLOSED (48/53 entities, 5 documented exemptions); `ogar-vocab-contract-codebook-migration-v1` CLOSED | `ogar-ar-shape-endgame-v1` Inc5 "F5-real" (one executor pair doing a REAL write) is the DEFERRED half that keeps the doctrine's §10 litmus at CONJECTURE | +| **Weather / OSINT / 3DGS / OCR** | weather D-WXS-0 Blocked on a classid mint, rest Queued; 22 3DGS docs untracked by design; D-OCR-51/52/53 unstarted | `substrate-comfort-zones-v1` CLOSED (hypothesis REFUTED — a result); weather POC v2 current | the `tesseract-rs-*` plans read OPEN here only because the real progress lives in the sibling repo's own `CLAUDE.md` (image→text byte-parity CLOSED there) — cross-repo status is not on this board | +| **Temporal / Markov / styles** (`temporal-markov-and-style-classes-v1`, `dtsc1-thinkingstyle-dedup-spec-v1`, `triangle-tenants-gestalt-separation-v1`, `thinking-engine-harvest-closure-v1`) | D-MTS-1/2/3 Queued — so the VSA→stream migration the 2026-07-10 ruling gates has not started; D-TSC-2/3, D-TRI-2..5 Queued; W0 census done, 51-file fate assignment not started | D-MTS-6 GREEN (k\*=1), D-TSC-1 resolved (M9), D-TRI-6 In PR | `VISION.md` §8 road (D-MTS-1 → D-TTV-1 → D-MTS-6b → …) is at its **first rung** | + +## 3. `.claude/v3/` — waves, ledger, open items (21 files, all read) + +| Wave | Verdict | Evidence | +|---|---|---| +| W0 ratify/document | **CLOSED** | D-V3-W0a/b Shipped (STATUS_BOARD:894-895); the folder exists as promised | +| W1 envelope + ownership (the keystone) | **CLOSED mechanism / PARTIAL adoption** | batch writer, delegation cache, probes SHIPPED (Addendum-15, INTEGRATION-PLAN:730-761); a real production caller exists (`cycle_driver.rs:516`, D-MBX-A6-P3c/P4) — but the board's own honesty ledger (STATUS_BOARD:1665): actor-owned production wiring NOT proven, durability FAKE until `LanceShardSink` | +| W2 kanban executors | **PARTIAL / CONTRADICTORY** | arm #1 Shipped (D-MBX-A6); W2a stated two ways (Addendum-12a vs -15, §1b); W2b's actor half DELETED 2026-08-05 — the wave-table row would send a session to rebuild the retired path | +| W3 compiled templates | **PARTIAL, mostly OPEN** | W3a `StepMask` shipped; W3b/c/d Queued; `compiled-templates.md` said StepMask did not exist (struck) | +| W4 DTO ladder | **PARTIAL** | D-PERT-1 Shipped (#630); W4a "In PR" as of 2026-07-10 (not re-verified in source); W4b Queued | +| W5 consumer adoption | **OPEN** | nine sub-items, none Shipped; smb-office-rs `LanceConnector::upsert` still the orphan write (W5f) | +| W6 monitor + retirement | **OPEN, premise RESCINDED** | D-CCF-4 rescinded 2026-07-03; W6a counting logic shipped, Lance-dataset sweep residue; "post-P4" now means post-checkpoint (⊘ notes landed) | + +`ENTROPY-MILESTONES.md` (M1–M27 after this PR): shipped/resolved M2 (regraded +here), M5, M6, M7, M9 (partially reopened 2026-07-18), M15, M25, M26; in-flight +M3, M4, M12, M13, M17, M20, M24 (regraded here); queued M8, M10, M11, M14, M16, +M19, M21, M22, M23, M27 (mechanism shipped, callers queued); **M18 RULING-NEEDED** +(sigma chain Ω→Δ→Φ→Θ→Λ vs `KanbanColumn` six phases — no ruling found). + +`knowledge/persona-vs-rung-ladder.md` O1–O9: **all open** except O5 (a probe +run; its two hardenings unconfirmed). O2 re-scoped 2026-08-30 to "any-rung +orchestration → `recipes::Recipe`"; O9 regraded StyleFamily's rung-4 anchoring +as a scalar-era artifact. `VISION.md` §8's dependency road (D-MTS-1 → D-TTV-1 → +D-MTS-6b → D-MTS-2/3 → D-TSC-2/3 → "the compilation loop closes") has **none of +its first three rungs shipped**. + +## 4. Harvest check — `.claude/v3/` did NOT have it; nine mirrors landed + +"Harvest" carries five senses in this workspace (ruff SPO producer; the +domain corpora it produces; cross-repo pattern-transfer briefs; the +literature/idea sweeps; and a same-word collision — see §7). The V3 folder +integrates sense 1 first-class (`knowledge/multi-anchor-ast-resolution.md`, +README row) and cited **neither** `.claude/nexgen/` (2026-09-05, four reader +reports + seven PR sweeps + the mask-histogram plan) **nor** +`.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` — although +`NestedBands` is keyed `(classid, version, idx)` (the V3 keyspace) and the +nexgen plan's room 27 proposes it as the carrier primer §6 demotes +`Vsa16kF32` from. Every "nexgen" string inside `.claude/v3/` meant the +`openproject-nexgen-rs` consumer. + +Landed in this PR: README doc-map row + collision note + two rulings added to +the canonical list; primer §5 ⊘ (D-CCF-4) + §6 row +(`E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1`, which postdated every +mirror in the folder); INTEGRATION-PLAN W6 ⊘; routing.md §5 ⊘; +compiled-templates `StepMask` struck; ENTROPY M2/M24 regrades + M27; +COMPONENT-MAP §6 `NestedBands` row; FUTURE-DESIGN 2026-09-07 block; +witness-nibble-lane P5 (lit-harvest D3: "a single `u8:u8` rail read as ONE +scalar axis is d=1", `:104`); the nexgen plan links back. **Deliberately not +added:** a `tenants.md` row — no D-NXG consumer sits in a `ValueTenant` yet, +and a row without a lane is a wish (ENTROPY's own meta-rule). + +## 5. W-waves + jc — verified against code (ndarray `master` `b9afcb9b`) + +| Item | State | Note | +|---|---|---| +| W1a #1 `from_i4_packed_u64` / `lane_i8` / `batch_packed_i4_16` | SHIPPED ×4 backends (scalar/avx512/neon + wasm partial) | TD said Open — fixed | +| W1a #2 `saturating_abs` (I8x16/I8x32) | SHIPPED ×4 | TD said Open — fixed | +| W1a #3 `gather_u16` / `palette_lookup_u8x8` | SHIPPED — **API only**; x86 body is a scalar-loop polyfill by its own doc (`simd_avx512.rs:2833-2836`) | performance intent unmet | +| W1a #4 `prefetch_read_t0/t1/t2` | SHIPPED ×3 | TD said Open — fixed | +| W1a #5 `U64x8::popcnt` / `xor_popcount` / `U64x4::popcnt` | SHIPPED (scalar/avx512/avx2) | TD said Open — fixed | +| W1b holograph `hamming.rs` | Open, 25 raw lines | — | +| W1b blasgraph | `types.rs` **0** (done, own path) / `ndarray_bridge.rs` **66** incl. the exact `_mm512_popcnt_epi64` #5 replaces | TD regraded half-done | +| W1b bgz17 `simd.rs` + `prefetch.rs` | Open, 22 + 1 | both gathers still raw; aarch64 `_prefetch` already a no-op | +| W1b contract `mul.rs::i4_eval` | Open, **64** (P0 TD-SIMD-SWEEP-W4) | untouched by #1/#2 | +| W1b thinking-engine | Open, 1 (`is_x86_feature_detected!`, `engine.rs:508`) | matvec already via `simd_amx` | +| W1.5 #6 `signature_pde_sweep` | SHIPPED (PR #293), consumed by `sigker::signature_kernel_pde` | — | +| W1.5 #7 `randomized_signature_sweep` | SHIPPED (PR #294), **consumed** (`sigker/src/randomized.rs:49,124`) | TD said Deferred — fixed | +| W1.5 #8 Lyndon pack | unbuilt, **ungated** (scalar prerequisite `sigker/src/log_signature.rs`, #1150); not a lane op by shape | TD note added | + +Raw-intrinsic census across the five named crates: **179 lines** (the +knowledge doc's "158-violation finding" of 2026-05-16 is stale as a live +number — drift from growth, not remediation; `types.rs` is the one real +reduction). **W1b verdict: 0 of 6 closed** with all five primitives shipped. + +jc: the 12-entry registry is literally "11/12 implemented, Pillar 2 deferred" +(`lib.rs:24-29`) — with the unstated caveat that Pillar 11 (`hambly_lyons`) +returns `PillarResult::deferred` under the crate's own `default = []` +(`hambly_lyons.rs:737-745`; `Cargo.toml:26,30`), so a plain `prove_it` shows +**10 executing + 2 deferred**. There is **no Pillar 6** in jc (1,2,3,4,5,5b, +7,8,9,9b,10,11); the board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) name a +different document's pillars. `jc::solver_order` is a 13th battery outside +`run_all_pillars()`. Four example files are present with no `[[example]]` +entry (`l9_loci_real_text`, `partof_isa_vs_palette256`, +`rung_divergence_reliability`, `weather_substrate_reliability`). W5's +PowerSig trigger is legitimately unfired: longest in-tree 4609 points vs +memory-half ≈ 11 585 (`√(2³⁰/8)`, recomputed) and time-half ≈ 33 388 (prior +run, not re-run here). + +## 6. Top-level `.claude/*.md` + dashboards + +- **71 files**; `CLAUDE.md:762` says 61 and `:765` cites `SESSION_CAPSTONE.md`, + which does not exist. **9 orphans** (zero refs, no marker, not in + CLAUDE.md's lists): `CALIBRATION_REPORT_2026_04_03`, the five-file + `LANGGRAPH_*` cluster (a coherent Python→Rust parity doc-set), + `WIKIDATA_EXTRACTION_PLAN`, `probe_m1_result_2026_04_11`, + `prompt-for-other-session-additive`. 61/71 predate 2026-06-01; only + `BOOT.md` (62 refs) and `ATTENTION_MASK_AUDIT_2026_08_21.md` are recent. + `pattern.md` (17 refs) and `patterns.md` (30 refs) are two live entry points + to one SoA/DTO doctrine. +- **STATUS_BOARD.md**: 1854 lines, 89 sections, 592 rows in 3/4/6-column + shapes; 427 cells carry narrative rather than a lifecycle token — a per-row + mechanical census is not possible without normalizing the table shape. +- **TECH_DEBT.md**: 167 entries under three header conventions; 65 unlabeled + (all pre-Kanban-format, `TD-F10-ACTOR-ID`..`TD-INT-14`); open P0s after + this PR: `TD-SIMD-SWEEP-W4`, `TD-API-DRIFT-MIDFLIGHT-1`, + `TD-SDR-PR-FOLLOWUP-1`, `TD-SDR-CONSUMER-PUSH-1`, `TD-OGIT-G-SLOT-1`. +- **ISSUES.md**: 47 entries, 37 OPEN. The ones this inventory touches: + `ISS-PLAN-TRACKING-IS-UNENFORCED`, `ISS-KANBAN-PLAN-EXIT-HAS-NO-NAMED-ROUTE`, + `ISS-RUNG-VS-BAND-CARDINALITY-COLLISION`, `ISS-REASONING-BAND-GATES-NOTHING`, + `ISS-EXCLUDED-CRATES-UNBUILT`, `ISS-SUPERSESSION-GENERATOR-DID-RANGE-NARROW`. + `ISS-CLASSID-OGAR-DRIFT` has two dated sub-entries (OPEN and RESOLVING). +- **SUPERSESSION-INDEX.md**: 14 ruled symbols, 73 blind plans (55 RESCOPE / + 18 READ / 0 ARCHIVE?); `BindSpace` RETIRE with 68 crate files / 47 plans / + 41 blind. + +## 7. Name collisions a session must read past + +| Word | Meaning A | Meaning B | +|---|---|---| +| **v3** | `.claude/v3/` mailbox-kanban-facet substrate | version 3 of the *retracted* `epistemic_bassin` 24-axis basis (`0x0334`) in `nexgen/harvest/14-*` and the literature harvest | +| **nexgen** | `.claude/nexgen/` (2026-09-05 harvest + plan) | `openproject-nexgen-rs` (consumer repo) — the only sense inside `.claude/v3/` until this PR | +| **Pillar 6 / Pillar 11** | jc registry: no Pillar 6; Pillar 11 = Hambly–Lyons uniqueness | board rows "Pillar-6/7" = another document's numbering; ndarray `hpc::pillar::signature` = a second "Pillar 11" (kernel stability) | +| **W1 … W5** | SIMD waves W1a/W1b/W1.5 (`ndarray-vertical-simd-alien-magic.md`) | `pillar11-signature-certification-unification-v1` W0–W6 (W1/W4 live in `ndarray/crates/sigker-parity`) | +| **harvest** | five senses (§4) | — | +| **EWA** | three different operations share the word (lit-harvest #27; jc sandwich = rendering push-forward) | — | +| **TrustTexture / GateDecision** | four `TrustTexture` types; two `GateDecision` types (M15 resolved the planner one) | `causal-edge/src/layout.rs` rules against a cast between them | + +## 8. Synergy — the digest (full text: EPIPHANIES 2026-09-07 block) + +Every open blocker in the SPOG/alpha/medcare/DataFusion arc is **read-time +re-derivation of something decided once upstream**: the ontology category +re-read from `value[96]` per row (medcare `obo_store.rs:16-18,77,681`), +`Domain::of_row` (`orphanet.rs:165`, `rails.rs:432`), `FacetRegime::PerRowTui` +(`domain_block.rs:259,695`), `graph_of(addr)` resolved per claim by a linear +scan (`spog_tenants.rs:38,87`), V1 external-id identity, and Lance row identity +re-materialised by `with_row_id`/`with_row_addr` (medcare `state.rs:900,936`). +The operator's sentence — *"every domain is just another table keyed by CUI +STDID (snomed) loinc etc — its a chain effect with masking, no datafusion +joins ever"* — is the schema-level form of the mask plan's invariant +(`substrate == mask geometry == projection surface`): one table per G, the +quad's 4×24 slots = four pre-resolved foreign keys, a crosswalk = a chain of +`eq_u32_to_mask` sweeps conjoined by `mask_ternlog` where the mask-out of hop +n is the key set of hop n+1 — never a mask-AND across two tables (nexgen room +18's family-separation rule is the mechanical "no joins"). The measured bounds +ride with it: masks win on Boolean relations at every density (D-GTM-0j — a +TYPE boundary), 0 bytes/step (0k), chaining pays only while masks stay in L2 +and survivors stay above 0.1 % (0n), and packed prefixes do NOT carry +cross-domain relations (0l) — the FK columns do. The seven harvest sources +collapse onto one object: the version-keyed `NestedBands` (+ overlap matrix) +per `(classid, version)` — reveal-ahead / cache-key = Lance version; Shannon = +WHERE only and it lags the popcount budget; EWA ranks WHERE to look, never +WHAT is true; a known unknown = a hop whose survivor mask has popcount > 1; +Boolean → masks, valued → blasgraph semirings over CSR. And the fence: rung +(processing horizon, 0–9) × ReasoningBand (assertion permission, bits 61..63) +× PlasticityState (how content changes, 3-bit S/P/O) are three axes; folding +any two into one "level" field is a LAYOUT-BREAK-class defect. + +## 9. What to do next (an order, not a plan — each step names its gate) + +1. **SPOG-aware bakes — one artifact per G** (medcare): the domain becomes + the table, `value[96]` stops being a read. Gate: PROBE-CROSSWALK-MASK-1 + (IDEAS 2026-09-07) — survivor set == the DataFusion join's set on one + fixture; 0 bytes/step; cross-family AND rejected at the seal. +2. **V3 lane-local addresses** — retire V1 external-id identity + (`E-A-V3-MINT-MUST-NEVER-DEGRADE-TO-V1-1` already bangs on the mint side). + Gate: D-TEMPORAL-1a's V3-accessor pin stays green. +3. **`ogar-r2il` gets a consumer** through `lance-graph-ogar` (`RANK` + + `TERNLOG 0x86`); today it has none. Gate: a lifted crosswalk program + equals the mask chain bit-for-bit; attaches to D-R2IL-5, not a new owner. +4. **DataFusion containment** — `with_row_id = false`, `with_row_addr = false` + pinned by a test that fails if either flips; no new DataFusion surface + (`E-PLANNING-MIGRATES-TO-LOCO-R2IL-DATAFUSION-IS-GRACE-PERIOD-1`). +5. Board hygiene the inventory could not do alone: adjudicate W2a + (Addendum-12a vs -15) from `canonical_node.rs:1622+`; rule M18; decide + `ISS-KANBAN-PLAN-EXIT-HAS-NO-NAMED-ROUTE`; retrofit the 65 unlabeled + TECH_DEBT rows or declare them open by policy; normalise STATUS_BOARD's + three table shapes so a mechanical census becomes possible. diff --git a/.claude/board/STATUS_BOARD.md b/.claude/board/STATUS_BOARD.md index 923c0781b..016becb4d 100644 --- a/.claude/board/STATUS_BOARD.md +++ b/.claude/board/STATUS_BOARD.md @@ -16,8 +16,8 @@ identity) is real and probe-gated. | D-LNC-4 | Mask convergence probe: lgj `MaskWords` (envelope rows 0..N) ≡ lance row-address prefilter bitmap. MUST show the identity holds on an append-only single-fragment fixture AND breaks on the first delete — one half is vacuous | Queued (independent of the bump) | | D-LNC-5 | Replace `VersionedGraph::diff`'s two-version full materialization (`read_all_batches` = `scan().try_into_stream()`, `versioned.rs:645`) with lance 11's native "row ids deleted between two versions". zero-copy-warden LENS-CLEAN; byte-identical output | Queued (blocked on D-LNC-3 — the API is lance 11) | | D-LNC-6 | Session reuse: end `Dataset::open`-per-call (17 sites in `lance-graph`); instrumented open-count strictly decreases, results identical | Queued (independent; lance 10 ships shared session) | -| D-LNC-5a | **Delta version-columns probe** (pre-registered, `crates/lance-graph/tests/delta_version_columns_probe.rs`): are `_row_created_at_version` / `_row_last_updated_at_version` populated WITHOUT stable row ids? Stable-row-id arm is the CONTROL so a null is attributable to lance, not the apparatus | **In PR** #1198. **RED, both arms.** insert: control 2, physical **0**; update: control HELD (version advanced AND value changed), stable **1**, physical **0**. ⊘ the update arm's first run was INCONCLUSIVE — its control did not hold because `MergeInsertBuilder::when_matched` defaults to `DoNothing`, so no update ever committed (caught by a CodeRabbit finding on #1198). **Consequence: D-LNC-5's insert/update half is gated behind the stable-row-id decision, same as the delete arm** | -| D-MW-P2 | **MemWAL crash-atomicity probe** (`crates/lance-graph/tests/memwal_atomicity_probe.rs`): the one measurement `.claude/temporal/08-memwal-vs-batchwriter.md` named as overturning its FOLD verdict — SIGKILL mid-flush on a 5 000-row put | **In PR** #1198. **GREEN.** A1 clean-drop control 5000; A2 SIGKILL sweep 5–500 ms → `0,0,0,5000,5000,5000,5000`. Boundary bracketed on both sides, no partial batch. FOLD stands. Scope recorded as NOT settled: one put, local store, no concurrent writers, no fencing path, and NOT the landing-rows-plus-frame-row composite (that is `LanceCycleWriter`'s, which has zero callers) | +| D-LNC-5a | **Delta version-columns probe** (pre-registered, `crates/lance-graph/tests/delta_version_columns_probe.rs`): are `_row_created_at_version` / `_row_last_updated_at_version` populated WITHOUT stable row ids? Stable-row-id arm is the CONTROL so a null is attributable to lance, not the apparatus | ~~**In PR** #1198~~ **Shipped** (#1198 merged `3797237b`; cell regraded 2026-09-07 plan inventory — it lagged the merge). **RED, both arms.** insert: control 2, physical **0**; update: control HELD (version advanced AND value changed), stable **1**, physical **0**. ⊘ the update arm's first run was INCONCLUSIVE — its control did not hold because `MergeInsertBuilder::when_matched` defaults to `DoNothing`, so no update ever committed (caught by a CodeRabbit finding on #1198). **Consequence: D-LNC-5's insert/update half is gated behind the stable-row-id decision, same as the delete arm** | +| D-MW-P2 | **MemWAL crash-atomicity probe** (`crates/lance-graph/tests/memwal_atomicity_probe.rs`): the one measurement `.claude/temporal/08-memwal-vs-batchwriter.md` named as overturning its FOLD verdict — SIGKILL mid-flush on a 5 000-row put | ~~**In PR** #1198~~ **Shipped** (#1198 merged `3797237b`; cell regraded 2026-09-07 plan inventory). **GREEN.** A1 clean-drop control 5000; A2 SIGKILL sweep 5–500 ms → `0,0,0,5000,5000,5000,5000`. Boundary bracketed on both sides, no partial batch. FOLD stands. Scope recorded as NOT settled: one put, local store, no concurrent writers, no fencing path, and NOT the landing-rows-plus-frame-row composite (that is `LanceCycleWriter`'s, which has zero callers) | | D-GTM-0n / P3 | **Ternlog chaining amortization** (`ndarray/examples/ternlog_amortization_probe.rs`, re-run `--release` 2026-09-06): does chaining pay on THIS mask shape? Pre-registered pass criterion = per-constraint cost flat in K, with a **K=1 control** (nothing to chain ⇒ no win) | **GREEN, bounded.** `T3/T1` bottoms at **0.50** by K=8; T3 total flat in K (57.6 ns @ K=4 → 54.1 ns @ K=32, 16× more constraints). K=1 control reads **1.03** as it must. Two bounds ride with the number and may not be dropped when it is cited: the win is **L2-residency-contingent** (bandwidth `138→15 GB/s` past L2, ratio back to 1.03 at a 512 KiB mask), and mask loses to sparse below **0.1%** active. **Does NOT retire the standing caveat** — 2× on one of three passes still cannot account for 5×. Gates Stage-3 SPEED claims only; the shape/correctness case never depended on it | | D-ALPHA-1b | **`AlphaMask::zip` release-mode length guard** (`lance-graph-contract/src/alpha.rs`): the only guard was `debug_assert_eq!`, compiled out in release, so mismatched-length operands produced a mask claiming `self.len` addresses over the SHORTER operand's word count — `count`/`is_empty` under-report silently, `contains`/`materialize_ordinals` panic at a distance | **Fixed** — promoted to `assert_eq!`. Defect reproduced in release BEFORE the fix (the `should_panic` falsifier reported "test did not panic as expected"); two-sided falsifiers both disable-verified. No API change, no external callers. Stage 1b of `.claude/temporal/09-plan.md`, now DONE | | D-LNC-7 | **One-WAL ruling**: `LanceCycleWriter`'s watermark WAL vs lance MemWAL — `applied_through` ≡ index catch-up position; two under one commit = two truths. Plan recommends keep-ours until D-LNC-3, then re-decide | **Operator decision** | diff --git a/.claude/board/TECH_DEBT.md b/.claude/board/TECH_DEBT.md index 3252cf44a..19d9a2998 100644 --- a/.claude/board/TECH_DEBT.md +++ b/.claude/board/TECH_DEBT.md @@ -1759,7 +1759,7 @@ multi-antecedent path for real. - **Severity:** P1 (blocks mul.rs follow-up + future i4-packed codec consumers) - **Surfaced in:** `simd-savant` PRE-MERGE audit 2026-05-16; PP-14 convergence-architect §SYNERGY 2 (see `.claude/knowledge/ndarray-vertical-simd-alien-magic.md`) -- **Status:** Open +- **Status:** ~~Open~~ **SHIPPED (primitive side).** Regraded 2026-09-07 (plan inventory, `PLAN-INVENTORY-2026-09-07.md` §W-waves): every symbol below exists on ndarray `master` (`b9afcb9b`) — `I8x16::from_i4_packed_u64` at `ndarray/src/simd_scalar.rs:1684` / `simd_avx512.rs:2673` / `simd_neon.rs:2269`, `batch_packed_i4_16` at `simd_scalar.rs:1971` / `simd_avx512.rs:3147` / `simd_neon.rs:2505`. This row had read "Open" for the whole time the symbols were merged. The CONSUMER half (`mul.rs::i4_eval::batch`, TD-SIMD-SWEEP-W4) is still Open — 64 raw-intrinsic lines untouched. - **Description:** `ndarray::simd` exposes `I8x16` / `I8x32` / `I8x64` typed wrappers and `dot_i8` / `min_i8` / `max_i8` / `add_i8` slice ops, but has no primitive for "unpack 16 signed nibbles from a `u64` into `I8x16` with sign-extension" — exactly the operation `crates/lance-graph-contract/src/mul.rs::i4_eval::batch` needs for any `QualiaI4_16D(u64)`-packed batch dispatch. PR #398 worked around it by inlining raw `_mm512_*` and `vld1q_u64` intrinsics (AP-SIMD-1/2 violations). - **Required API surface (file as parallel PR against `adaworldapi/ndarray` master):** - `impl I8x16 { pub fn from_i4_packed_u64(packed: u64) -> Self; }` — AVX-512 via `_mm512_cvtepi8_epi16` + nibble shuffle; NEON via `vshl_n_s8`; scalar fused-loop fallback. @@ -1773,7 +1773,7 @@ multi-antecedent path for real. - **Severity:** P1 (closes codex P2 i8::MIN divergence on PR #398 by giving consumers a single source-of-truth for hardware-semantics abs) - **Surfaced in:** PR #398 codex P2 review; PP-16 preflight-drift-auditor verdict "Direction B" 2026-05-16 -- **Status:** Open +- **Status:** ~~Open~~ **SHIPPED (primitive side).** Regraded 2026-09-07 (plan inventory): `saturating_abs` on `I8x16`/`I8x32` at `ndarray/src/simd_scalar.rs:1709,1737` / `simd_avx512.rs:2716,2937` / `simd_neon.rs:2301,2325`. Consumer half (TD-SIMD-SWEEP-W4) still Open. - **Description:** Scalar path in `mul.rs` uses `signed_mantissa.unsigned_abs() as i8`, which wraps `i8::MIN = -128` back to `-128i8` (the cast `u8 → i8` doesn't saturate), then `-128 ≤ 1` is true → wrongly classifies as `ValleyOfDespair`. PR #398's AVX-512 path correctly classifies `i8::MIN` not because of VPABSB (VPABSB does NOT saturate — `abs(0x80) = 0x80`, the bit pattern is unchanged), but because the path widens i8 → i64 first and then negate-blends, where the negate of -128 (i64) is +128 (i64), comparing > 1. Spec line 233 of `pr-sprint-13-simd-i4.md`: `|signed_mantissa| ≤ 1 → ValleyOfDespair` represents weak rule signal, NOT sign-extreme. Direction B (scalar is buggy, AVX-512 outcome is correct) is canonical — but the new ndarray primitive must produce truly-saturating semantics across all three backends. - **Required API surface:** - `impl I8x16 { pub fn saturating_abs(self) -> Self; }` — AVX-512 `_mm512_min_epu8(_mm512_abs_epi8(x), _mm512_set1_epi8(0x7f))` (VPABSB leaves `0x80 → 0x80`; VPMINUB clamps `0x80` unsigned-greater-than `0x7f` down to `0x7f`); NEON `vqabsq_s8` (the `q` suffix is hardware-saturating); scalar `i8::saturating_abs` fused loop. @@ -1787,7 +1787,7 @@ multi-antecedent path for real. - **Severity:** P1 (blocks `bgz17/src/simd.rs` migration off raw `_mm256_i32gather_epi32`) - **Surfaced in:** `simd-savant` PRE-MERGE audit 2026-05-16 (location: `crates/bgz17/src/simd.rs:88`) -- **Status:** Open +- **Status:** ~~Open~~ **SHIPPED — API only, PARTIAL on intent.** Regraded 2026-09-07 (plan inventory): `U16x8::gather_u16` at `ndarray/src/simd_scalar.rs:1799` / `simd_avx512.rs:2846` / `simd_neon.rs:2356`, `palette_lookup_u8x8` at `simd_scalar.rs:1863` / `simd_avx512.rs:2900` / `simd_neon.rs:2423`. The x86_64 body is, by its own doc comment (`simd_avx512.rs:2833-2836`), "a scalar-loop polyfill (real AVX2 gather via `_mm256_i32gather_epi32` + downcast is tracked as a follow-up optimisation)" — so the contract (bounds-safe, three backends agree) is met and the performance intent of this row is not. Consumer half (`bgz17/src/simd.rs`, TD-SIMD-SWEEP-W3) still Open and still calls both gathers raw. - **Description:** `bgz17` uses `_mm256_i32gather_epi32` directly for palette lookup (8 u16 indices → values). `ndarray::simd` exposes no gather primitive — the polyfill needs `U16x8::gather_u16` or a dedicated `palette_lookup_u8x8` helper. Dominant palette-stream workload primitive; missing it forces every palette consumer to reinvent gather. - **Required API surface:** - `impl U16x8 { pub fn gather_u16(indices: U16x8, table: &[u16]) -> Self; }` — AVX2 `_mm256_i32gather_epi32` + downcast; NEON scalar loop (no native gather); scalar `indices.iter().map(|&i| table[i])`. @@ -1800,7 +1800,7 @@ multi-antecedent path for real. - **Severity:** P2 (perf-only; closes 2 AP-SIMD-1 violations in `bgz17/src/prefetch.rs`) - **Surfaced in:** `simd-savant` PRE-MERGE audit 2026-05-16 (locations: `crates/bgz17/src/prefetch.rs:96` x86 `_mm_prefetch`, `:100` aarch64 `_prefetch`) -- **Status:** Open +- **Status:** ~~Open~~ **SHIPPED (primitive side).** Regraded 2026-09-07 (plan inventory): `prefetch_read_t0` at `ndarray/src/simd_scalar.rs:1887` / `simd_avx512.rs:3091` / `simd_neon.rs:2445` (t1/t2 adjacent). Consumer half (TD-SIMD-SWEEP-W3) still Open: `prefetch.rs:96` x86 `_mm_prefetch` remains; the aarch64 `_prefetch` cited above is already gone (replaced by a documented no-op, `prefetch.rs:98-108`). - **Description:** `bgz17/prefetch.rs` issues `_mm_prefetch` (x86) and `_prefetch` (aarch64) directly. `ndarray::simd` has no prefetch hint API — the polyfill needs cross-arch `prefetch_read_t0` / `_t1` / `_t2` helpers that no-op on unsupported targets. - **Required API surface:** - `pub fn prefetch_read_t0(ptr: *const u8);` — AVX `_mm_prefetch(_, _MM_HINT_T0)`; NEON `__builtin_prefetch(_, 0, 3)`-equivalent; no-op on unsupported. (Or wrap `core::intrinsics::prefetch_read_data` once stable.) @@ -1814,7 +1814,7 @@ multi-antecedent path for real. - **Severity:** P1 (blocks holograph/hamming + blasgraph hamming migration off raw `_mm512_popcnt_epi64`) - **Surfaced in:** `simd-savant` PRE-MERGE audit 2026-05-16 (locations: `crates/holograph/src/hamming.rs:530,567,637,638`; `crates/lance-graph/src/graph/blasgraph/types.rs:440,484`; `crates/lance-graph/src/graph/blasgraph/ndarray_bridge.rs:245`) -- **Status:** Open +- **Status:** ~~Open~~ **SHIPPED (primitive side).** Regraded 2026-09-07 (plan inventory): `U64x8::popcnt` / `xor_popcount` at `ndarray/src/simd_scalar.rs:1916,1934` / `simd_avx512.rs:2987,3046`, `U64x4::popcnt` at `simd_scalar.rs:1952`. Consumer halves (TD-SIMD-SWEEP-W1 holograph, W2 `ndarray_bridge.rs`) still Open — `ndarray_bridge.rs:465,493` still call `_mm512_popcnt_epi64` raw, the exact call this primitive exists to replace. - **Description:** `holograph/hamming.rs` and `lance-graph/blasgraph/types.rs` use `_mm512_popcnt_epi64` for 64-bit lane popcounts on AVX-512 VPOPCNTDQ. `ndarray::hpc::bitwise::popcount_raw` covers the slice case (already exposed) but there is no `ndarray::simd::U64x8::popcnt()` lane-wise method. Consumers fall back to raw intrinsics. - **Required API surface:** - `impl U64x8 { pub fn popcnt(self) -> Self; }` — AVX-512 `_mm512_popcnt_epi64` (VPOPCNTDQ); NEON `vcntq_u8` + horizontal-sum; scalar `u64::count_ones` fused loop. @@ -1838,11 +1838,11 @@ multi-antecedent path for real. --- -### TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (W1.5-#7, DEFERRED) +### TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (W1.5-#7, SHIPPED) - **Severity:** P3 (deferred; gated on `jc Pillar 11`) - **Surfaced in:** sigker architectural review 2026-05-16 -- **Status:** Deferred +- **Status:** ~~Deferred~~ **SHIPPED (ndarray PR #294) and CONSUMED.** Regraded 2026-09-07 (plan inventory): `ndarray/src/hpc/randomized_signature.rs` — `INCREMENT_EPSILON` (`:96`, mirrored from sigker), `randomized_signature_step` (`:205`), `randomized_signature_sweep_with` (`:244`), `randomized_signature_sweep` (`:292`); consumed by sigker — the import `use ndarray::hpc::randomized_signature::randomized_signature_sweep` at `crates/sigker/src/randomized.rs:49`; the call site is `RandomizedSignatureBuilder::encode` (`randomized.rs:124`). Three-way inconsistency at the time of the regrade: `.claude/knowledge/ndarray-vertical-simd-alien-magic.md` and ndarray's `vertical-simd-consumer-contract.md` both said SHIPPED since 2026-09-02, this row said Deferred. Shipped as `f64`/`Vec` k×k GEMV + axpy per increment (the sigker carrier), not the `F32x16::random_proj_step` sketch below — the same carrier correction #6 records. - **Description:** sigker's randomized signatures (Cuchiero-Schmocker-Teichmann 2021 universality) are fixed-width finite-dim projections of the path signature. The hot path is a Gaussian-random-matrix-vector update with `F32x16` state — same shape as W1a-#1 closure-batch primitive, different lane type. - **Required API surface (when activated):** - `impl F32x16 { pub fn random_proj_step(state: Self, seed: u64, depth: u32) -> Self; }` — single-step Gaussian projection update. @@ -1855,7 +1855,7 @@ multi-antecedent path for real. - **Severity:** P3 (deferred; gated on `jc Pillar 11`) - **Surfaced in:** sigker architectural review 2026-05-16 -- **Status:** Deferred +- **Status:** Deferred — **no longer gated** (note added 2026-09-07, plan inventory): the W1.5 gate opened 2026-09-02 and the scalar prerequisite now exists (`crates/sigker/src/log_signature.rs`, lance-graph #1150: `enumerate_lyndon_words` / `log_signature_truncated` / `witt_dimension`). Still unbuilt in ndarray (0 hits for `lyndon_pack`), and by shape it is a table-driven pack over a Lyndon index, not a lane op (`.claude/nexgen/harvest/04-sigker-power-kernels.md`) — build it as a scalar-anchored primitive if a consumer appears, never as a SIMD claim. The "7-13×" below is the N ≥ 8 regime only: measured 2.1× (d=4, N=2) → 7.6× (N=8) → 11.6× (N=12), `crates/sigker/src/lib.rs`. - **Description:** Log-signatures compress the truncated signature into the Lyndon basis of the free Lie algebra (7-13× compression, lossless). The pack/unpack primitives operate on `I16x16` state with combinatorial-index awareness. - **Required API surface (when activated):** - `impl I16x16 { pub fn lyndon_pack(self, basis_idx: u8) -> Self; }` @@ -1878,7 +1878,7 @@ multi-antecedent path for real. - **Severity:** P2 (Open; gated on W1a-#5 + retire `U8x64::nibble_popcount_lut` duplicate) - **Surfaced in:** `simd-savant` audit 2026-05-16 -- **Status:** Open +- **Status:** Open — **HALF DONE** (regraded 2026-09-07, plan inventory): `types.rs` carries 0 raw-intrinsic lines — `hamming_distance_dispatch` routes through `ndarray::hpc::bitwise::hamming_distance_raw` (`types.rs:462`), migrated on its own path. `ndarray_bridge.rs` is unchanged: 66 raw lines, `_mm512_popcnt_epi64` at `:465` and `:493` where `U64x8::popcnt`/`xor_popcount` (W1a-#5, shipped) should be consumed. The "22 raw ops" for `types.rs` below is history. - **Description:** `crates/lance-graph/src/graph/blasgraph/types.rs` (22 raw ops) + `ndarray_bridge.rs` (60 raw ops, plus 2 AP-SIMD-7 duplicate LUTs) — Hamming + popcount workload over palette CSR/CSC. Migration consumes `U64x8::popcnt`, existing `U8x64::nibble_popcount_lut`, and `ndarray::hpc::bitwise::hamming_distance_raw`. - **Cross-ref:** `crates/lance-graph/src/graph/blasgraph/{types.rs:440-506, ndarray_bridge.rs:149-299}`; TD-NDARRAY-SIMD-POPCOUNT-U64. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md new file mode 100644 index 000000000..9258ea7a4 --- /dev/null +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md @@ -0,0 +1,336 @@ +# Plans inventory A-K (first half, alphabetical, 1..103) + +Repo: `/home/user/lance-graph` at `aeebfb23` (main). Range: `2026-05-06-splat-osint-ingestion-v1.md` .. `kognitionswirtschaft-v1.md` (103 files, confirmed via `ls .claude/plans | sort` entries 1-103). + +## Table + +| file | title | status_token | version | d_ids | prs | last_commit | verdict | +|---|---|---|---|---|---|---|---| +| 2026-05-06-splat-osint-ingestion-v1.md | Splat-OSINT Ingestion v1 — SPLAT-1 stage 0->1 + EWA OSINT bridge | Active | v1 | D-SPLAT-1, D-SPLAT-2, D-SPLAT-3, D-SPLAT-4, D-SPLAT-5, D-SPLAT-6, D-SPLAT-7 | #286, #287, #289, #322, #323, #324 | 2026-05-06 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| 3DGS-3D-Tiles-runtime-plan.md | 3DGS 3D Tiles Runtime Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-4x4-cognitive-shader-integration-plan.md | 3DGS 4x4 Cognitive-Shader Integration Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-ArcGIS-Cesium-ingestion-plan.md | 3DGS ArcGIS and Cesium Ingestion Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-Blender-transcode-crosspollination-plan.md | 3DGS Blender Transcode Cross-Pollination Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-Cesium-BindSpace4-headstone-exploration.md | 3DGS Cesium / BindSpace4 Headstone Exploration — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-Cesium-feature-mapping-plan.md | 3DGS Cesium Feature Mapping Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-HHTL-datalake-traversal-plan.md | 3DGS HHTL Datalake Traversal Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-Lance-Arrow-storage-plan.md | 3DGS Lance/Arrow Storage Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-PLAN-INDEX.md | 3DGS Implementation Plan Index — lance-graph | NONE | - | D-T | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-PRX12-crosspollination-capstone.md | 3DGS / PR-X12 Cross-Pollination Capstone — lance-graph | NONE | - | D-PHASE, D-WHP | #696, #697 | 2026-07-16 | **AMBIGUOUS** — no status line found | +| 3DGS-SplatShaderBlas-BLASGraph-crosspollination-plan.md | 3DGS SplatShaderBlas / BLASGraph Cross-Pollination Plan — lance-graph | NONE | - | - | - | 2026-07-16 | **AMBIGUOUS** — no status line found | +| 3DGS-blast-radius-application-map.md | 3DGS Blast Radius Application Map — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-certified-query-render-plan.md | 3DGS Certified Query and Render Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-cross-pollination-raw-field-plan.md | 3DGS Cross-Pollination Raw-Field Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-domain-adapter-strategy-plan.md | 3DGS Domain Adapter Strategy Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-epiphany-roadmap-plan.md | 3DGS Epiphany Roadmap Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-genetics-4x4-fanout-plan.md | 3DGS Genetics 4x4 Fanout Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-integration-wiring-plan.md | 3DGS Integration Wiring Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-neuronal-network-4x4-plan.md | 3DGS Neuronal Network 4x4 Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| 3DGS-ultrasound-SaMD-plan.md | 3DGS Ultrasound SaMD Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| Palette256-3DSB-PhiSpiral-attention-integration-plan.md | Palette256 / 3DSB / PhiSpiral Attention Integration Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| PhiSpiral256-SoA-cross-system-integration-plan.md | PhiSpiral256 SoA Cross-System Integration Plan — lance-graph | NONE | - | - | - | 2026-05-25 | **AMBIGUOUS** — no status line found | +| SYNERGY-MAP-S00-S07.md | SYNERGY-MAP-S00-S07 — reuse / transcode / wire, not build | INSPIRATION / CONJECTURE | - | D-1, D-GR, D-MBX-A6, D-MTS-1, D-MTS-5, D-ONLY, D-PALETTE-NNUE, D-QUANTGATE | #210, #211, #227 | 2026-08-01 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| a3-carrier-v1.md | A3 — the `I4x32D` carrier + `pack`/`unpack` (plan v1, 2026-06-01) | NONE | v1 | D-A3-1, D-A3-2, D-A3-3, D-A3-4, D-A3-5, D-A3-6, D-ATOM-1 | - | 2026-06-01 | **AMBIGUOUS** — no status line found | +| alpha-channel-rung-overlay-v1.md | Alpha-channel rung overlay — v1 | PROPOSED | v1 | D-1, D-ACR-0, D-ACR-1, D-ACR-10, D-ACR-11, D-ACR-12, D-ACR-13, D-ACR-14, D-ACR-15, D-ACR-16, D-ACR-17, D-ACR-2, D-ACR-3, D-ACR-4, D-ACR-5, D-ACR-6, D-ACR-7, D-ACR-8, D-ACR-9, D-AWARENESS-1, D-BLW-3, D-CV3-3, D-HTT-6, D-LEG-REASONING-LIVES-IN-LANCE, D-OR-IT-IS-…(+1 more, 26 total) | #444, #971, #973, #978 | 2026-08-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| alpha-interventional-faithfulness-v1.md | Plan: a recorded alpha is instrumentation until it is interventionally load-bearing (`alpha-interventional-faithfulness-v1`) | PROPOSAL | v1 | D-AIF-1, D-AIF-2, D-AIF-3, D-AIF-4, D-AIF-5, D-AIF-6, D-ECG-7 | #1055, #1057 | 2026-08-26 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| alpha-reason-witness-cognitive-fabric-v1.md | alpha-reason-witness-cognitive-fabric-v1 | PROPOSED / PLAN ONLY | v1 | D-ARW-0, D-ARW-1, D-ARW-2, D-ARW-3, D-ARW-4, D-ARW-5, D-ARW-6, D-ARW-7, D-ARW-8 | #1077, #879 | 2026-08-30 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| alpha-reason-witness-shader-field-archaeology-pass-1.md | D-ARW-0 shader-field archaeology — pass 1 | SOURCE AUDIT / PLAN ONLY | - | D-ARW-0 | #1051, #130 | 2026-08-30 | **CLOSED** — all 1 cited D-ids Shipped on board | +| alpha-reason-witness-shader-field-d-arw-0-audit-v1.md | D-ARW-0 shader-field source audit v1 | SOURCE AUDIT / PLAN-ONLY | v1 | D-ARW-0, D-ARW-8, D-SF-HINDSIGHT-1, D-SF-RUNG-1, D-SF-V3-3 | #1043, #1078 | 2026-08-30 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| alpha-reason-witness-shader-field-lineage-addendum-v1.md | alpha-reason-witness cognitive fabric — shader-field lineage addendum v1 | ADDENDUM TO #1078 / PLAN ONLY | v1 | D-ARW-0, D-ARW-1, D-ARW-4, D-ARW-5, D-ARW-6, D-ARW-8 | #1043, #1051, #1078 | 2026-08-30 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| anatomy-realtime-v1.md | anatomy-realtime-v1 — The Proof-of-Vision Plan | NONE | v1 | D-10, D-ANATOMY-DEMO-8, D-CASCADE-V1-11 | #289, #337, #346, #347, #355 | 2026-05-12 | **AMBIGUOUS** — no status line found | +| archetype-scaffold-v1.md | Archetype Transcode Crate Scaffold — v1 | In progress | v1 | D-AND-SCAFFOLDED, D-MAPPING-INCOMPLETE | - | 2026-04-24 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| atom-mailbox-substrate-v1.md | atom-mailbox-substrate-v1 | PROPOSAL | v1 | D-ATOM-0, D-ATOM-1, D-ATOM-2, D-ATOM-3, D-ATOM-4, D-ATOM-5, D-PERSONA-1, D-PERSONA-5 | - | 2026-05-27 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| belief-abi-restoration-v1.md | BELIEF-ABI-RESTORATION-1 — which Belief semantics still have no home? | CHARTER | v1 | D-A-MEMORY-ABI-ESCAPE-1, D-WITNESS-1 | #1000, #1004, #1007 | 2026-08-23 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| belief-abi-step2-addendum-copula-v1.md | BELIEF-ABI-RESTORATION-1 — Step 2 ADDENDUM: the copula correction | ADDENDUM to ` | v1 | - | - | 2026-08-23 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| belief-abi-step2-ruling-request-v1.md | BELIEF-ABI-RESTORATION-1 — Step 2: the ruling request | RULING REQUEST | v1 | D-POLARITY | #1006, #1007, #1009, #1010, #1011 | 2026-08-23 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| bindspace-columns-v1.md | BindSpace Columns E/F/G/H — Integration Plan v1 | Active | v1 | D-AWARENESS-INLINE-1, D-DIST-1, D-E1, D-E2, D-E3, D-E4, D-E5, D-E6, D-E7, D-F1, D-F10, D-F2, D-F3, D-F4, D-F5, D-F6, D-F7, D-F8, D-F9, D-G1, D-G2, D-G3, D-G4, D-G5, D-H1, D-H2, D-H3, D-H4 | #270 | 2026-04-27 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| bindspace-mailbox-soa-dependency-map-v1.md | bindspace → mailbox_soa — dependency map & two-path wiring preflight (v1) | MAP / preflight | v1 | D-MBX-A1, D-MBX-A2 | - | 2026-06-17 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| bindspace-mailbox-soa-w3-w4a-impl-v1.md | W3+W4a — atomic implementation plan (code-grounded, v1) | v2 | v1 | D-BLOCKED, D-DIST-5 | #517, #518, #519 | 2026-06-17 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| bindspace-mailbox-soa-wiring-v1.md | BindSpace → MailboxSoA — the WIRING plan (v1) | MEASURED / ready-to-execute | v1 | D-BSW-0, D-BSW-1, D-BSW-2, D-BSW-3, D-BSW-4, D-V3-W3, D-V3-W4, D-V3-W7 | #1174, #518, #844 | 2026-09-05 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| bindspace-singleton-to-mailbox-soa-v1.md | bindspace-singleton-to-mailbox-soa-v1 — dissolve the shared `Arc` into per-mailbox `MailboxSoA` thoughtspace | CONJECTURE / design | v1 | D-CE64-MB-1, D-CE64-MB-5, D-CSV-5, D-CSV-7, D-MBX-10, D-MBX-11, D-MBX-12, D-MBX-2, D-MBX-6, D-MBX-7, D-MBX-8, D-MBX-9, D-MBX-A1, D-MBX-A2, D-MBX-A3, D-MBX-A4, D-MBX-A5, D-MBX-A6, D-PERSONA-5, D-RECONCILE, D-RESONANCEDTO-DUP-1, D-SURREALDB-KVLANCE-LANCE7 | #386, #423, #433, #445, #477, #487, #490, #530, #533, #535, #537 | 2026-06-18 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| burn-ndarray-parity-sprint-v1.md | Burn ↔ ndarray Parity Sprint v1 | Active | v1 | - | #115 | 2026-04-30 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| callcenter-membrane-v1.md | Plan — External Callcenter Membrane (Supabase-shape over Lance + DataFusion) | Active | v1 | - | - | 2026-04-24 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cam-pq-production-wiring-v1.md | Plan — CAM-PQ Production Wiring (2026-04-20) | DRAFT | v1 | - | #218 | 2026-04-20 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| capstone-cognitive-loop-wiring-nan-census-v1.md | Capstone Validation Plan — Phase-Aligned Cognitive Loop: Wiring & NaN-Census (v1) | PROPOSED | v1 | D-BRAIN-IS-ZERO-COPY-IFF-FIXEDSIZEBINARY, D-COPY | #572 | 2026-06-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| capstone-out-leg-wiring-v1.md | Capstone OUT-leg wiring — closing S2 / S3 / S4 / run-NaN (v1) | PLAN | v1 | - | #571, #572, #573, #574, #578 | 2026-06-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cascade-seal-register-grid-v1.md | Cascade-accumulated seal on the register grid — RATIFIED v3 (5+3 council) | RATIFIED v3 | v1 | D-BY, D-SEAL-SPEC | #968 | 2026-08-20 | **AMBIGUOUS** — status says RATIFIED / "council complete" (design review finished, not implementation) but D-BY and D-SEAL-SPEC are NOT on STATUS_BOARD — cannot confirm shipped | +| categorical-algebraic-inference-v1.md | Categorical-Algebraic Inference — The Architecture That Clicks | Active | v1 | - | - | 2026-04-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| causal-rung-standing-wave-v1.md | Plan — causal rung trajectories + standing-wave awareness (p64→v3 cognition layer) | CONJECTURE throughout | v1 | D-AW-5, D-CSW-1, D-CSW-2, D-CSW-3, D-IMPLICIT-1, D-IN-SCOPE | #774, #776, #787 | 2026-08-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| causaledge64-mailbox-rename-soa-v1.md | CausalEdge64 Mailbox + Rename SoA — Integration Plan v1 | Active | v1 | D-CE64-MB-1, D-CE64-MB-2, D-CE64-MB-3, D-CE64-MB-4, D-CE64-MB-5, D-CE64-MB-6, D-CE64-MB-7, D-CE64-MB-8, D-CE64-MB-9, D-INT4-32D-ATOMS-6, D-OGIT-G-1, D-THINKING-ENGINE-UNWIRED-1 | #146, #355, #366, #369, #370 | 2026-05-14 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cesium-osm-substrate-v1.md | cesium-osm-substrate-v1 — integration plan (debt + future) | PROPOSAL / integration plan | v1 | D-HIGH, D-OSM, D-OSM-1, D-OSM-2, D-OSM-3, D-OSM-4, D-OSM-5, D-OSM-6, D-OSM-7, D-SPLAT-1, D-SPLAT-12, D-SPLAT-2, D-SPLAT-3, D-SPLAT-4, D-SPLAT-6, D-T | #123, #142, #471, #472 | 2026-06-05 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| classid-canon-custom-flip-v1.md | Classid Canon:Custom Half-Order Flip — Plan v1 (the migration is TRIGGERED) | ACTIVE | v1 | D-HUMANREADABLE-REORDER-DEFERRED, D-OSINT-APPID-NOT-CONCEPT, D-SPLIT-ORDER-IS-A-FLIP, D-VCW-4 | #146, #618, #626, #627 | 2026-07-02 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| codec-sweep-via-lab-infra-v1.md | Plan — Codec Candidate Sweep via Lab Infra (JIT-first, no rebuilds) | NONE | v1 | D-R | #12662, #219, #220, #221 | 2026-04-20 | **AMBIGUOUS** — no status line found | +| cognitive-compilation-v1.md | Cognitive Compilation for Lance-Graph — v1 | PROPOSED | v1 | D-CC-COMPILER-1, D-CC-EQUIV-1, D-CC-FENCE-1, D-CC-INDEX-1, D-CC-LEDGER-1, D-CC-OGAR-1, D-CC-PROMOTE-1, D-CC-REVIEW-1, D-CC-RIG-1, D-CC-RUBICON-1, D-CC-RUNTIME-1 | - | 2026-06-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cognitive-substrate-convergence-v1.md | Cognitive Substrate Convergence — i4 Mantissa, Gapless Baton, Active Inference (v1) | PROPOSAL | v1 (higher v3 exists) | D-ATOMS, D-CSV, D-CSV-1, D-CSV-10, D-CSV-11, D-CSV-12, D-CSV-2, D-CSV-3, D-CSV-4, D-CSV-5, D-CSV-6, D-CSV-7, D-CSV-8, D-CSV-9 | #116, #366, #372, #373, #374, #375, #376, #378, #379 | 2026-05-16 | **SUPERSEDED** — higher version sibling exists ([2, 3]) | +| cognitive-substrate-convergence-v2.md | Cognitive Substrate Convergence — i4 Mantissa, Gapless Baton, Active Inference (v2) | ACTIVE | v2 (higher v3 exists) | D-7, D-COLLAPSE-GATE-SMALLVEC-1, D-CSV, D-CSV-1, D-CSV-10, D-CSV-11, D-CSV-12, D-CSV-13, D-CSV-14, D-CSV-15, D-CSV-2, D-CSV-3, D-CSV-4, D-CSV-5, D-CSV-6, D-CSV-7, D-CSV-8, D-CSV-9, D-D-CSV-8-SIMD-1, D-SHADER-DRIVER-WORKSPACE-CONFLICT-1, D-SIGMA-TIER-THRESHOLDS…(+1 more, 22 total) | #116, #372, #373, #375, #379, #381, #383, #384, #385, #386, #387 | 2026-05-16 | **SUPERSEDED** — higher version sibling exists ([3]) | +| cognitive-substrate-convergence-v3.md | Cognitive Substrate Convergence — i4 Mantissa, Gapless Baton, Active Inference (v3) | ACTIVE | v3 | D-7, D-COLLAPSE-GATE-SMALLVEC-1, D-CSV, D-CSV-1, D-CSV-10, D-CSV-11, D-CSV-12, D-CSV-13, D-CSV-14, D-CSV-15, D-CSV-16, D-CSV-17, D-CSV-2, D-CSV-3, D-CSV-4, D-CSV-5, D-CSV-6, D-CSV-7, D-CSV-8, D-CSV-9, D-ISOLATION, D-SHADER-DRIVER-WORKSPACE-CONFLICT-1, D-SIGMA-…(+1 more, 24 total) | #147, #372, #381, #383, #384, #385, #386, #387, #388, #389, #391 | 2026-05-16 | **OPEN** — status ACTIVE; "complete" refers to sprint waves not whole plan; sprint-13 scope locked (ongoing); many D-ids not on board or only partially merged | +| cognitive-write-roundtrip-substrate-v1.md | Cognitive Write = Round-Trip Substrate Hardening — v1 | PLAN | v1 | D-BACK | - | 2026-06-09 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| compile-time-consumer-binding-v1.md | Compile-Time Consumer Binding — Tier-2 Sub-Plan (Patterns E + F) | Active | v1 | D-CIRCULAR-COMPILATION-7, D-CONTEXT-BUNDLE-2, D-GENERIC-BRIDGE-3, D-MANIFEST-MODULES, D-MANIFEST-MODULES-4, D-OGIT-G-SLOT-1, D-RACTOR-SUPERVISOR, D-RACTOR-SUPERVISOR-5 | - | 2026-05-12 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| counterfactual-rung3-closure-v1.md | Counterfactual rung-3 closure — encoding is specified, runtime is absent — v1 | DESIGN | v1 | D-1, D-2, D-3, D-ADDRESSING-REASSESSMENT, D-ATOM-4, D-CFR-1, D-CFR-2, D-CFR-3, D-CFR-4, D-CFR-5, D-CFR-6, D-CSW-1, D-LATER, D-MTS-6, D-TRI-5, D-TRI-6, D-TSC-1 | - | 2026-08-20 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cycle-coherent-soa-snapshot-v1.md | Plan: Cycle-Coherent SoA Snapshot — No-Cross-Cycle-Lag Guarantee | Queued | v1 | D-SOA-SNAP-1, D-SOA-SNAP-2, D-SOA-SNAP-3, D-SOA-SNAP-4, D-SOA-SNAP-5, D-SOA-SNAP-6 | #468, #477 | 2026-06-11 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cycle-loop-closure-driver-v1.md | cycle-loop-closure-driver-v1 — the loop-closure driver that makes the persist_sink cycle/WAL seam load-bearing | IMPLEMENTED | v1 | D-BLW-1, D-BLW-2, D-BLW-3, D-BLW-4, D-BLW-5, D-IGN-B, D-KIA-A2, D-LEG-REASONING-LIVES-IN-LANCE-GRAPH-1, D-MBX, D-MBX-9-IN, D-MBX-A1, D-MBX-A6-P1, D-MBX-A6-P3, D-MBX-A6-P4, D-ORDER, D-OUT, D-TRANSITION, D-V3-W2 | #879, #888 | 2026-09-05 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| cypher-kanban-ast-unification-v1.md | cypher-kanban-ast-unification-v1 — Cypher IS the kanban-board AST; the GUID-keyed substrate IS the graph; four subsystems collapse to one | PLAN | v1 | D-IS-THE-GRAPH | #540 | 2026-06-18 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dacr7-band-reading-contract-v1.md | SPEC v1 — D-ACR-7: the CE64 59..63 reading contract | NONE | v1 | D-ACR-0, D-ACR-1, D-ACR-17, D-ACR-2, D-ACR-7, D-AND-SPLIT | - | 2026-08-21 | **AMBIGUOUS** — no status line found | +| deepnsm-morton-comma-facet-v1.md | DeepNSM × Morton-comma facet — design + PROBE-DEEPNSM-FACET spec (v1) | SPEC | v1 | D-CONVERGENCE-1 | #173 | 2026-07-19 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| deepnsm-v3-convergence-v1.md | DeepNSM → V3 substrate convergence — v1 | PROPOSED | v1 | D-1, D-ARM-7, D-DNV-1, D-DNV-2, D-DNV-3, D-DNV-4 | #496, #500, #624, #626 | 2026-07-04 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dialectic-engine-v1.md | dialectic-engine v1 — the reasoning cathedral (synthesis of the operator's six pillars, three theses, two antitheses) | ACTIVE | v1 | D-AUTOCOMPLETE, D-CLOSE-TRANSITIVE-HORIZON-CAP, D-DEEPNSM-V2-BELIEF-DUP, D-ELEVATION-1, D-HHTL-TRIE-FIELD-SEARCH-1, D-LEG-REASONING-LIVES-IN-LANCE-GRAPH-1, D-SEARCH-LOOP-1, D-TILE256 | - | 2026-07-23 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dismech-causal-replay-v1.md | dismech-causal-replay-v1 — causality replay over the DisMech palette | PROPOSED | v1 | D-AND-LEFT-EMPTY-1, D-CV3, D-CV3-0, D-DCR, D-DCR-0, D-DCR-1, D-DCR-2, D-DCR-3, D-DCR-4, D-DCR-5, D-DCR-6, D-HOME-NEEDS-A-FIRST-CONSUMER-OR-IT-IS-A-VACANCY-1, D-MATH-1 | #1075, #1118, #1127, #879 | 2026-09-03 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dismech-causality-v3-v1.md | DisMech × Causality-V3 — repository-grounded rebase report, v1 | REPORT, no code | v1 | D-10-GM, D-ARM-1, D-CV3-0, D-CV3-1, D-CV3-2, D-CV3-3, D-CV3-4, D-CV3-5, D-CV3-6 | - | 2026-08-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dtri1-classid-mint-spec-v1.md | D-TRI-1 classid-half — batched OGAR mint spec — v1 | SPEC | v1 | D-GR-5, D-TRI, D-TRI-1, D-TRI-1-P4, D-TRI-4 | #717, #719, #722 | 2026-07-18 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| dtsc1-thinkingstyle-dedup-spec-v1.md | D-TSC-1 SPEC v1 — M9 ThinkingStyle dedup (council-grade, 5+3 protocol) | NONE | v1 | D-TSC-1, D-TSC-2, D-TSC-3 | - | 2026-07-10 | **AMBIGUOUS** — no status line found | +| elegant-herding-rocket-v1.md | Plan — DeepNSM as Full Parser via Markov ±5 Context Upgrade | NONE | v1 | - | #204, #208, #209 | 2026-04-19 | **AMBIGUOUS** — no status line found | +| entropy-closure-causal-ground-v1.md | Plan: entropy measures closure; bits 59–60 tell whether the closure has causal footing (`entropy-closure-causal-ground-v1`) | PROPOSAL | v1 | D-ACR-7, D-ECG-1, D-ECG-2, D-ECG-3, D-ECG-4, D-ECG-5, D-ECG-6, D-ECG-7 | #1057 | 2026-08-26 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| entropy-ladder-spo-rung-v1.md | Entropy-Ladder SPO Rung Decomposition — v1 | PLATEAU | v1 | D-EL-1, D-EL-2, D-EL-3, D-EL-4, D-EL-5, D-EL-6, D-EL-COCA | - | 2026-06-14 | **OPEN** — PLATEAU: only R1 shipped, R2-R6 explicitly "planned" — plan remains open, not naive-substring, phase-scoped SHIPPED | +| epiphany-integration-2026-07-04-v3.md | Epiphany Integration Plan — Membranes, Parity, and the Unified ruff Phase Sequence | NONE | v3 | D-1, D-BY-A4, D-BY-A5, D-BY-A6, D-BY-A7, D-BY-B1, D-BY-B3, D-COMPAT-READER, D-ON, D-OPACITY-1, D-REF-1, D-STATE-IS-A-RANGE-1, D-THINKING-TEMPLATES, D-TO-CONTRACT, D-TO-RBAC | #139, #147, #148, #151, #215, #620, #628, #645 | 2026-07-04 | **AMBIGUOUS** — no status line found | +| episodic-risc-spine-v1.md | Episodic RISC Spine — v1 (the converged episodic addressing architecture) | NONE | v1 | D-EW64-1, D-EW64-2, D-H2H-1, D-HORIZON-1, D-MBX-9-IN, D-STORY-1, D-STORY-2, D-STORY-3, D-VIEW-1 | #444 | 2026-05-31 | **AMBIGUOUS** — no status line found | +| epistemic-quadrant-materialization-v1.md | Plasticity materialization — grey/white as MECHANISM, over the B6 stance panel | PROPOSED | v1 | D-FORK-CANNOT-CLOSE-WHAT-SINGLES-CANNOT, D-PROJECTION, D-QUANTGATE, D-SCI-4, D-SRS-3, D-SRS-4 | #539, #629, #862 | 2026-07-29 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| erasure-seals-compaction-research-v1.md | erasure-seals-compaction-research-v1 — RP-SEAL (operator charter, 2026-08-18) | DISPATCHED 2026-08-18 | v1 | D-DV-IS-EPISTEMIC-1, D-REAL | #879, #911, #912, #913, #961 | 2026-08-19 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| ew64-witness-unification-v1.md | EW64 × CausalWitness — the witness demarcation and the EpisodicWitness64 landing zone — v1 | DESIGN / SEAM-CUT | v1 | D-1, D-2, D-3, D-4, D-5, D-ADDRESSING-REASSESSMENT, D-ATOM-4, D-ATOM-5, D-EWU-1, D-EWU-2, D-EWU-3, D-EWU-4, D-EWU-5, D-EWU-6, D-EWU-7, D-EWU-8, D-EWU-9, D-TSC-1 | #444 | 2026-08-20 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| foundry-consumer-parity-v1.md | Foundry Consumer Parity — Shared Ontology + Contracts for SMB + MedCare | Active | v1 | - | - | 2026-04-28 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| foundry-roadmap-unified-smb-medcare-v1.md | Foundry Roadmap — Unified SMB + MedCare Consumer Plan v1 | Active | v1 | D-10 | #276 | 2026-04-28 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| genetic-research-substrate-integration-v1.md | Genetic Research Substrate Integration v1 — implementation plan | initial proposal | v1 | D-COARSERESIDUE-NO-VALUE-TENANT, D-GEN-1, D-GEN-10, D-GEN-2, D-GEN-3, D-GEN-4, D-GEN-5, D-GEN-6, D-GEN-7, D-GEN-8, D-GEN-9 | #218, #491, #494, #495, #496, #498, #500 | 2026-06-16 | **OPEN** — FALSE-POSITIVE CAUGHT: statusline says "No code shipped yet" (negated) — naive substring match on "shipped" wrongly flagged closed; correct verdict is OPEN (initial proposal, 0/11 D-ids on board) | +| genetics-probes-v1.md | Genetics Substrate — Gating Probes v1 | PLANTED 2026-06-16 | v1 | D-GEN, D-GEN-1, D-GEN-10, D-GEN-2, D-GEN-3, D-GEN-7, D-GEN-CHAODA-ENSEMBLE | #217, #218, #219, #220, #500, #501 | 2026-06-16 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| golden-vs-tempered-stride-v1.md | golden-vs-tempered-stride-v1 — head vs gut, made falsifiable | T1–T4 RUN | v1 | - | #935 | 2026-08-12 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| grammar-foundry-followup-v1.md | Grammar + Foundry Follow-up — v1 | Active | v1 | - | #243, #275, #278, #280, #281, #283, #284, #285 | 2026-04-29 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| graph-desktop-platform-synergy-map-v1.md | Graph Desktop Platform — Synergy Map v1 | vocabulary | v1 | D-KV, D-MTS-1 | #210, #216, #217, #366, #763 | 2026-07-19 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| graphrag-doc-retrieval-soa-integration-v1.md | graphrag + doc-retrieval on the V3 SoA — integration plan v1 | DESIGN + FIRST CODE | v1 | D-1, D-GR-1, D-GR-2, D-GR-3, D-GR-4, D-GR-5, D-GR-6, D-KV, D-OF-COGNITION-1, D-OGAR-DOC-LAYER, D-TRI, D-TRI-1, D-TRI-6 | #631, #707, #708, #709, #720 | 2026-08-25 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| grounding-descent-cognitive-maslow-v1.md | grounding-descent-cognitive-maslow-v1 — when a high rung is handed a sloppy joe, descend the needs ladder | PROPOSED | v1 | D-1, D-ATOM-4, D-GDM-1, D-GDM-2, D-GDM-3, D-GDM-4, D-GDM-5, D-RUNG-MUL-1 | #969, #970 | 2026-08-26 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| guid-v2-tail-per-family-codebook-v1.md | Integration Plan — GUID v2 tail (3×u16) + per-family codebook scoping (v1) | PROPOSED | v1 | D-AMORT, D-FIELD-NOT-A-VALUE-TYPE, D-GV2-1, D-GV2-2, D-GV2-3, D-GV2-4, D-GV2-5 | - | 2026-06-20 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| h268-probe-wave-v1.md | H.268 Probe Wave v1 — WH-MAG · SIG-CHECKSUM · WALK-SPECTRUM | NONE | v1 | D-QUANTGATE | #242, #245, #695, #701 | 2026-07-16 | **AMBIGUOUS** — no status line found | +| hhtl-thinking-tables-le-contract-v1.md | HHTL thinking tables × the little-endian structural contract — v1 | RATIFIED | v1 | D-ADDRESSING-REASSESSMENT, D-BY, D-HTT-1, D-HTT-10, D-HTT-11, D-HTT-2, D-HTT-3, D-HTT-4, D-HTT-5, D-HTT-6, D-HTT-7, D-HTT-8, D-HTT-9, D-RUNG2-144-VOCAB-SPLIT | #249 | 2026-08-20 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| house-differential-style-v1.md | House differential style v1 — a differential-diagnosis style program over shipped rung-3 tactics | D-HOUSE-1 REPORTED 2026-09-02 | v1 | D-1, D-ATOM-1, D-DCR-3, D-HOUSE-0, D-HOUSE-1, D-HOUSE-2, D-HOUSE-3, D-HOUSE-4, D-HOUSE-5, D-HOUSE-6, D-PERSONA-1, D-TEH-2, D-TSC-2 | #1136, #1137 | 2026-09-02 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| hydrate-crate-hardening-council-v1.md | 5+3 Council: hardening `lance-graph-hydrate` before merge (PR #957) | NONE | v1 | - | #907, #957 | 2026-08-17 | **AMBIGUOUS** — no status line found | +| identity-architecture-exists-vs-needs-v1.md | Identity Architecture — What Exists vs What Needs Building (v1) | INTEGRATION MAP + PLAN | v1 | D-CLASSID-WIDTH, D-CLS, D-IDENTITY-1, D-IDENTITY-2 | #480, #484, #489, #490 | 2026-06-13 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| idle-flush-dataset-eviction-v1.md | Idle-flush dataset eviction — plan v1 | PROPOSAL | v1 | - | #901, #907 | 2026-08-07 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| integrated-cognitive-planner-v1.md | Integrated Cognitive Planner — meticulous reference map (v1, 2026-06-15) | NONE | v1 | D-EL-2, D-MBX-A6-P3, D-ORDER, D-RESONANCEDTO-DUP-1, D-SURREALDB-KVLANCE-LANCE7, D-VALUESCHEMA-FULL-POC-DEFAULT | #223, #372, #386, #437, #439, #489, #490, #492, #495, #496 | 2026-06-15 | **AMBIGUOUS** — no status line found | +| integration-actionhandler-rbac-orchestration-v1.md | Integration Plan — ActionHandler ⟷ RBAC ⟷ Orchestration spine (v1) | NONE | v1 | - | - | 2026-06-23 | **AMBIGUOUS** — no status line found | +| jc-pillars-runtime-wiring-v1-ERRATUM.md | Erratum to `jc-pillars-runtime-wiring-v1.md` | NONE | - | - | - | 2026-05-05 | **AMBIGUOUS** — no status line found | +| jc-pillars-runtime-wiring-v1.md | JC pillars → runtime wiring (cognitive-shader-driver) | plan, not implementation | v1 | - | - | 2026-05-05 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| kanban-64k-inverted-awareness-v1.md | kanban-64k-inverted-awareness v1 — parallel thinking + the inverted-awareness witness | PLANNED / CONJECTURE-graded per section | v1 | D-KIA-0, D-KIA-A1, D-KIA-A2, D-KIA-B1, D-KIA-C5, D-KIA-D1, D-KIA-D3, D-MBX-A6-P3, D-STATS-DEGENERACY-CONTRACT-DIVERGENCE | #879 | 2026-08-04 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| known-unknown-handover-network-v1.md | Known-unknown handover network — v1 | PROPOSED | v1 | D-1, D-ACR-1, D-ACR-10, D-ACR-11, D-ACR-15, D-ACR-16, D-ACR-4, D-ACR-5, D-ACR-7, D-ACR-8, D-CV3-0, D-CV3-3, D-KUH-1, D-KUH-2, D-KUH-3, D-KUH-4, D-SINK-ANTIPATTERN-1 | - | 2026-08-21 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | +| kognitionswirtschaft-v1.md | Kognitionswirtschaft v1 — Maslow · Heckhausen/Rubikon · Libet as one cognitive economy | PROPOSED | v1 | D-1, D-ACR-8, D-KW-1, D-KW-2, D-KW-3, D-KW-4, D-KW-5, D-KW-6 | - | 2026-08-26 | **OPEN** — status line indicates active/proposed/plan/design/queued (default) | + +## Counts + +- OPEN: 64 +- CLOSED: 1 +- SUPERSEDED: 2 +- AMBIGUOUS: 36 +- **Total: 103**## Mismatches + +Per task instructions, a "mismatch" includes both (a) a D-id whose board +status contradicts the plan's own claim, and (b) a D-id the plan cites that +has no row in `STATUS_BOARD.md` at all. Type (a) contradictions were +specifically hunted for (see Candidate epiphanies below) and only weak/ +inferential cases turned up — none where a plan explicitly claims a D-id is +shipped/complete while the board explicitly says Queued/In progress for that +same D-id. Type (b) is common and is the bulk of this section. + +**314 total D-id citations, across 61 plans, have NO row in STATUS_BOARD.md.** + +### Plans where NONE of the cited D-ids appear on STATUS_BOARD.md (26 files) + +- `3DGS-PLAN-INDEX.md`: D-T +- `3DGS-PRX12-crosspollination-capstone.md`: D-PHASE, D-WHP +- `a3-carrier-v1.md`: D-A3-1, D-A3-2, D-A3-3, D-A3-4, D-A3-5, D-A3-6, D-ATOM-1 +- `anatomy-realtime-v1.md`: D-10, D-ANATOMY-DEMO-8, D-CASCADE-V1-11 +- `archetype-scaffold-v1.md`: D-AND-SCAFFOLDED, D-MAPPING-INCOMPLETE +- `belief-abi-restoration-v1.md`: D-A-MEMORY-ABI-ESCAPE-1, D-WITNESS-1 +- `belief-abi-step2-ruling-request-v1.md`: D-POLARITY +- `bindspace-columns-v1.md`: D-AWARENESS-INLINE-1, D-DIST-1, D-E1, D-E2, D-E3, D-E4, D-E5, D-E6, D-E7, D-F1, D-F10, D-F2, D-F3, D-F4, D-F5, D-F6, D-F7, D-F8, D-F9, D-G1, D-G2, D-G3, D-G4, D-G5, D-H1, D-H2, D-H3, D-H4 +- `bindspace-mailbox-soa-w3-w4a-impl-v1.md`: D-BLOCKED, D-DIST-5 +- `capstone-cognitive-loop-wiring-nan-census-v1.md`: D-BRAIN-IS-ZERO-COPY-IFF-FIXEDSIZEBINARY, D-COPY +- `cascade-seal-register-grid-v1.md`: D-BY, D-SEAL-SPEC +- `codec-sweep-via-lab-infra-v1.md`: D-R +- `cognitive-write-roundtrip-substrate-v1.md`: D-BACK +- `compile-time-consumer-binding-v1.md`: D-CIRCULAR-COMPILATION-7, D-CONTEXT-BUNDLE-2, D-GENERIC-BRIDGE-3, D-MANIFEST-MODULES, D-MANIFEST-MODULES-4, D-OGIT-G-SLOT-1, D-RACTOR-SUPERVISOR, D-RACTOR-SUPERVISOR-5 +- `cycle-coherent-soa-snapshot-v1.md`: D-SOA-SNAP-1, D-SOA-SNAP-2, D-SOA-SNAP-3, D-SOA-SNAP-4, D-SOA-SNAP-5, D-SOA-SNAP-6 +- `cypher-kanban-ast-unification-v1.md`: D-IS-THE-GRAPH +- `deepnsm-morton-comma-facet-v1.md`: D-CONVERGENCE-1 +- `dialectic-engine-v1.md`: D-AUTOCOMPLETE, D-CLOSE-TRANSITIVE-HORIZON-CAP, D-DEEPNSM-V2-BELIEF-DUP, D-ELEVATION-1, D-HHTL-TRIE-FIELD-SEARCH-1, D-LEG-REASONING-LIVES-IN-LANCE-GRAPH-1, D-SEARCH-LOOP-1, D-TILE256 +- `epiphany-integration-2026-07-04-v3.md`: D-1, D-BY-A4, D-BY-A5, D-BY-A6, D-BY-A7, D-BY-B1, D-BY-B3, D-COMPAT-READER, D-ON, D-OPACITY-1, D-REF-1, D-STATE-IS-A-RANGE-1, D-THINKING-TEMPLATES, D-TO-CONTRACT, D-TO-RBAC +- `erasure-seals-compaction-research-v1.md`: D-DV-IS-EPISTEMIC-1, D-REAL +- `foundry-roadmap-unified-smb-medcare-v1.md`: D-10 +- `genetic-research-substrate-integration-v1.md`: D-COARSERESIDUE-NO-VALUE-TENANT, D-GEN-1, D-GEN-10, D-GEN-2, D-GEN-3, D-GEN-4, D-GEN-5, D-GEN-6, D-GEN-7, D-GEN-8, D-GEN-9 +- `genetics-probes-v1.md`: D-GEN, D-GEN-1, D-GEN-10, D-GEN-2, D-GEN-3, D-GEN-7, D-GEN-CHAODA-ENSEMBLE +- `grounding-descent-cognitive-maslow-v1.md`: D-1, D-ATOM-4, D-GDM-1, D-GDM-2, D-GDM-3, D-GDM-4, D-GDM-5, D-RUNG-MUL-1 +- `guid-v2-tail-per-family-codebook-v1.md`: D-AMORT, D-FIELD-NOT-A-VALUE-TYPE, D-GV2-1, D-GV2-2, D-GV2-3, D-GV2-4, D-GV2-5 +- `h268-probe-wave-v1.md`: D-QUANTGATE + +### Plans with PARTIAL board coverage — the D-ids below are cited but absent from the board (the plan's other D-ids DO have board rows; see the main table) + +- `SYNERGY-MAP-S00-S07.md` (3/8 found): missing → D-1, D-GR, D-ONLY, D-PALETTE-NNUE, D-QUANTGATE +- `alpha-channel-rung-overlay-v1.md` (12/26 found): missing → D-1, D-ACR-10, D-ACR-11, D-ACR-12, D-ACR-13, D-ACR-14, D-ACR-15, D-ACR-16, D-ACR-17, D-AWARENESS-1, D-HTT-6, D-LEG-REASONING-LIVES-IN-LANCE, D-OR-IT-IS-A-JOKE-1, D-SINK-ANTIPATTERN-1 +- `alpha-reason-witness-cognitive-fabric-v1.md` (1/9 found): missing → D-ARW-1, D-ARW-2, D-ARW-3, D-ARW-4, D-ARW-5, D-ARW-6, D-ARW-7, D-ARW-8 +- `alpha-reason-witness-shader-field-d-arw-0-audit-v1.md` (1/5 found): missing → D-ARW-8, D-SF-HINDSIGHT-1, D-SF-RUNG-1, D-SF-V3-3 +- `alpha-reason-witness-shader-field-lineage-addendum-v1.md` (1/6 found): missing → D-ARW-1, D-ARW-4, D-ARW-5, D-ARW-6, D-ARW-8 +- `atom-mailbox-substrate-v1.md` (2/8 found): missing → D-ATOM-0, D-ATOM-1, D-ATOM-2, D-ATOM-3, D-ATOM-4, D-ATOM-5 +- `bindspace-mailbox-soa-wiring-v1.md` (5/8 found): missing → D-V3-W3, D-V3-W4, D-V3-W7 +- `bindspace-singleton-to-mailbox-soa-v1.md` (18/22 found): missing → D-CSV-5, D-RECONCILE, D-RESONANCEDTO-DUP-1, D-SURREALDB-KVLANCE-LANCE7 +- `causal-rung-standing-wave-v1.md` (4/6 found): missing → D-IMPLICIT-1, D-IN-SCOPE +- `causaledge64-mailbox-rename-soa-v1.md` (7/12 found): missing → D-CE64-MB-8, D-CE64-MB-9, D-INT4-32D-ATOMS-6, D-OGIT-G-1, D-THINKING-ENGINE-UNWIRED-1 +- `cesium-osm-substrate-v1.md` (13/16 found): missing → D-HIGH, D-OSM, D-T +- `classid-canon-custom-flip-v1.md` (1/4 found): missing → D-HUMANREADABLE-REORDER-DEFERRED, D-OSINT-APPID-NOT-CONCEPT, D-SPLIT-ORDER-IS-A-FLIP +- `cognitive-compilation-v1.md` (6/11 found): missing → D-CC-FENCE-1, D-CC-INDEX-1, D-CC-LEDGER-1, D-CC-PROMOTE-1, D-CC-REVIEW-1 +- `cognitive-substrate-convergence-v1.md` (9/14 found): missing → D-ATOMS, D-CSV, D-CSV-11, D-CSV-5, D-CSV-6 +- `cognitive-substrate-convergence-v2.md` (12/22 found): missing → D-7, D-COLLAPSE-GATE-SMALLVEC-1, D-CSV, D-CSV-11, D-CSV-5, D-CSV-6, D-D-CSV-8-SIMD-1, D-SHADER-DRIVER-WORKSPACE-CONFLICT-1, D-SIGMA-TIER-THRESHOLDS-1, D-TRUST-TEXTURE-DUPE-1 +- `cognitive-substrate-convergence-v3.md` (14/24 found): missing → D-7, D-COLLAPSE-GATE-SMALLVEC-1, D-CSV, D-CSV-11, D-CSV-5, D-CSV-6, D-ISOLATION, D-SHADER-DRIVER-WORKSPACE-CONFLICT-1, D-SIGMA-TIER-THRESHOLDS-1, D-TRUST-TEXTURE-DUPE-1 +- `counterfactual-rung3-closure-v1.md` (6/17 found): missing → D-1, D-2, D-3, D-ADDRESSING-REASSESSMENT, D-ATOM-4, D-CFR-2, D-CFR-3, D-CFR-4, D-CFR-5, D-CFR-6, D-LATER +- `cycle-loop-closure-driver-v1.md` (11/18 found): missing → D-LEG-REASONING-LIVES-IN-LANCE-GRAPH-1, D-MBX, D-MBX-A6-P3, D-ORDER, D-OUT, D-TRANSITION, D-V3-W2 +- `dacr7-band-reading-contract-v1.md` (4/6 found): missing → D-ACR-17, D-AND-SPLIT +- `deepnsm-v3-convergence-v1.md` (5/6 found): missing → D-1 +- `dismech-causal-replay-v1.md` (8/13 found): missing → D-AND-LEFT-EMPTY-1, D-CV3, D-DCR, D-HOME-NEEDS-A-FIRST-CONSUMER-OR-IT-IS-A-VACANCY-1, D-MATH-1 +- `dismech-causality-v3-v1.md` (8/9 found): missing → D-10-GM +- `dtri1-classid-mint-spec-v1.md` (3/5 found): missing → D-TRI, D-TRI-1-P4 +- `episodic-risc-spine-v1.md` (5/9 found): missing → D-HORIZON-1, D-STORY-1, D-STORY-2, D-STORY-3 +- `epistemic-quadrant-materialization-v1.md` (2/6 found): missing → D-FORK-CANNOT-CLOSE-WHAT-SINGLES-CANNOT, D-PROJECTION, D-QUANTGATE, D-SCI-4 +- `ew64-witness-unification-v1.md` (2/18 found): missing → D-1, D-2, D-3, D-4, D-5, D-ADDRESSING-REASSESSMENT, D-ATOM-4, D-ATOM-5, D-EWU-2, D-EWU-3, D-EWU-4, D-EWU-5, D-EWU-6, D-EWU-7, D-EWU-8, D-EWU-9 +- `graph-desktop-platform-synergy-map-v1.md` (1/2 found): missing → D-KV +- `graphrag-doc-retrieval-soa-integration-v1.md` (7/13 found): missing → D-1, D-GR-3, D-KV, D-OF-COGNITION-1, D-OGAR-DOC-LAYER, D-TRI +- `hhtl-thinking-tables-le-contract-v1.md` (1/14 found): missing → D-ADDRESSING-REASSESSMENT, D-BY, D-HTT-10, D-HTT-11, D-HTT-2, D-HTT-3, D-HTT-4, D-HTT-5, D-HTT-6, D-HTT-7, D-HTT-8, D-HTT-9, D-RUNG2-144-VOCAB-SPLIT +- `house-differential-style-v1.md` (11/13 found): missing → D-1, D-ATOM-1 +- `identity-architecture-exists-vs-needs-v1.md` (2/4 found): missing → D-CLASSID-WIDTH, D-CLS +- `integrated-cognitive-planner-v1.md` (1/6 found): missing → D-MBX-A6-P3, D-ORDER, D-RESONANCEDTO-DUP-1, D-SURREALDB-KVLANCE-LANCE7, D-VALUESCHEMA-FULL-POC-DEFAULT +- `kanban-64k-inverted-awareness-v1.md` (6/9 found): missing → D-KIA-D3, D-MBX-A6-P3, D-STATS-DEGENERACY-CONTRACT-DIVERGENCE +- `known-unknown-handover-network-v1.md` (11/17 found): missing → D-1, D-ACR-10, D-ACR-11, D-ACR-15, D-ACR-16, D-SINK-ANTIPATTERN-1 +- `kognitionswirtschaft-v1.md` (7/8 found): missing → D-1## Candidate epiphanies + +Facts only, counts and filenames included; no speculation. + +1. **35/103 plans (34%) have no `Status:` line anywhere in the document** — + `status_token = NONE` for all of them. Two distinct sub-populations: + - **22 are the "3DGS design-doc" family** (`3DGS-*`, `Palette256-3DSB-*`, + `PhiSpiral256-SoA-*`) — pure architecture/mapping documents with no + `Status:` line, essentially no D-ids (only `3DGS-PLAN-INDEX.md` and + `3DGS-PRX12-crosspollination-capstone.md` cite any), and mostly no PR + citations. They read as a different genre from the rest of `.claude/plans/` + — narrative/reference docs, not tracked delivery plans — and are the + reason `3DGS-PLAN-INDEX.md` exists as their own index. + - **13 others** (`a3-carrier-v1.md`, `anatomy-realtime-v1.md`, + `codec-sweep-via-lab-infra-v1.md`, `dacr7-band-reading-contract-v1.md`, + `dtsc1-thinkingstyle-dedup-spec-v1.md`, `elegant-herding-rocket-v1.md`, + `epiphany-integration-2026-07-04-v3.md`, `episodic-risc-spine-v1.md`, + `h268-probe-wave-v1.md`, `hydrate-crate-hardening-council-v1.md`, + `integrated-cognitive-planner-v1.md`, + `integration-actionhandler-rbac-orchestration-v1.md`, + `jc-pillars-runtime-wiring-v1-ERRATUM.md`) DO look like normal tracked + plans (several cite D-ids and PRs — e.g. `a3-carrier-v1.md` cites 7 + D-ids and `dtsc1-thinkingstyle-dedup-spec-v1.md` cites 3) but simply + never got a `Status:` header. This is a real board-hygiene gap distinct + from the 3DGS-family case. + +2. **314 of 559 D-id citations (56%) across these 103 plans have NO row in + `.claude/board/STATUS_BOARD.md`.** 61/103 plans (59%) cite at least one + untracked D-id; 26 plans have ZERO board coverage for any D-id they cite + (full list in Mismatches). This directly instantiates the known, + already-named issue the board itself documents: + `STATUS_BOARD.md`'s own bindspace-mailbox-soa-wiring-v1 section says + "a board-only mint is invisible to `supersession_index.py`'s coverage + column, which is the root cause `ISS-PLAN-TRACKING-IS-UNENFORCED` (#1156) + names." This inventory is fresh, independent measurement confirming that + the gap is large and current, not historical. + +3. **Only one plan family in this half has multiple versions:** + `cognitive-substrate-convergence-v{1,2,3}.md`. Neither v1 nor v2 marks + itself SUPERSEDED, ⊘, or otherwise self-declares obsolescence — v1's own + status line says "PROPOSAL (sprint-10 architectural decisions + consolidated...)" and v2's says "ACTIVE — sprint-11 Phase A/B COMPLETE...". + Supersession is inferable ONLY from the existence of the newer sibling + file, confirming the task brief's warning that self-declared status text + cannot be trusted to signal supersession — the mechanical version-sibling + check is doing real work here, independent of prose. + +4. **Two live, freshly-caught instances of the exact naive-substring trap + `CLAUDE.md` warns about** (a status word appearing inside text that + does NOT predicate the whole plan): + - `genetic-research-substrate-integration-v1.md` — status line reads + "...10 deliverables (D-GEN-1..10) + 3 gating probes. **No code shipped + yet.**" A substring match on "shipped" would misclassify this CLOSED; + the word is negated. Correct reading: OPEN (0/11 of its D-ids are on + the board at all). + - `entropy-ladder-spo-rung-v1.md` — status line reads "PLATEAU — R1 + (foundation) **SHIPPED**; R2–R6 planned (probe-gated where marked)." + SHIPPED describes one sub-phase (R1) of a six-phase plan, not the + whole document. A substring match would misclassify CLOSED; correct + reading is OPEN (R2-R6 remain). + - Related but softer: `cognitive-substrate-convergence-v3.md` — "Wave F + + Wave G **complete** (grade B+); sprint-13 entry scope locked" reads + as an in-progress multi-sprint plan describing a completed phase, not + a finished plan (also several of its D-ids are `NOT-FOUND` or only + partially merged on the board). + +5. **`cascade-seal-register-grid-v1.md` status line explicitly says + "RATIFIED v3 — council complete (Phases 0–5)"** — the strongest closure- + sounding language of any file in this batch — yet BOTH of its D-ids + (`D-BY`, `D-SEAL-SPEC`) are completely absent from `STATUS_BOARD.md`. + "RATIFIED" here evidently means the design/spec review process finished, + not that an implementation shipped; the board gives no way to confirm + implementation status. Verdict recorded as AMBIGUOUS, not CLOSED. + +6. **`D-ARW-0` is cited by 4 different plans** (`alpha-reason-witness- + cognitive-fabric-v1.md`, `-shader-field-archaeology-pass-1.md`, + `-shader-field-d-arw-0-audit-v1.md`, `-shader-field-lineage-addendum- + v1.md`) and shows `Shipped` on the board, while each of those 4 plans' + own status line says "PLAN ONLY" / "no production implementation in + this file" / "SOURCE AUDIT". This is consistent (the shipped work landed + outside these particular audit/plan documents) but is a real plan↔board + split-of-labor pattern worth knowing before trusting a single plan file + to reflect a D-id's true status. + +7. **PR #879 is cited by 5 different plans** in this range + (`cycle-loop-closure-driver-v1.md`, `dismech-causal-replay-v1.md`, + `erasure-seals-compaction-research-v1.md`, + `kanban-64k-inverted-awareness-v1.md`, and one more) — the most-cited PR + number in this half of the alphabet. #386 and #372 are each cited by 4 + plans. A session investigating any of these PRs should expect to find + its scope spread across multiple plan documents. + +8. **`git log` shows every one of the 20 `3DGS-*` + `Palette256-3DSB-*` + + `PhiSpiral256-SoA-*` design docs sharing the exact same last-commit date + `2026-05-25`**, except `3DGS-PRX12-crosspollination-capstone.md` and + `3DGS-SplatShaderBlas-BLASGraph-crosspollination-plan.md` (both + `2026-07-16`). This is consistent with the whole design-doc family being + authored/landed in one batch commit, reinforcing finding #1's reading + that they are a single coherent genre, not independently tracked plans.## Unread / uncertain + +All 103 files were read (metadata region: title line, first grep-located +status line + surrounding context where ambiguous, full D-id/PR scan across +the whole file). Files below needed extra manual verification beyond the +mechanical pass, or remain genuinely uncertain after it: + +- `cascade-seal-register-grid-v1.md` — read in full context around its + "RATIFIED v3 — council complete" status line specifically to check whether + "complete" meant implementation-complete or review-complete; concluded the + latter (design/spec ratification) because neither of its 2 D-ids appears + on `STATUS_BOARD.md` — but this is an inference from absence, not a + positive statement anywhere that implementation is still pending. Recorded + AMBIGUOUS rather than CLOSED or OPEN. +- `hhtl-thinking-tables-le-contract-v1.md` — carries an in-body + "⊘ SUPERSEDED IN PART" note (line 3) pointing at + `docs/architecture/ARC-B-OWNERSHIP-AND-ADDRESSING-REASSESSMENT.md` for 4 + specific corrected claims, while its own `Status:` line says "RATIFIED + (operator review, 2026-08-19)". Treated as a live, ratified plan with + in-place corrections (OPEN), NOT as a superseded plan — "in part" plus an + explicit RATIFIED status argues against whole-document SUPERSEDED — but + this is a judgment call a reader with more context on ARC-B might weigh + differently. +- `deepnsm-morton-comma-facet-v1.md` — similarly has an in-body + "⊘ SUPERSEDED" note (line 108) that retires one specific measurement + clause (§3.5's "3b" set-membership test), not the plan. Its own status + line ("SPEC (design handoff for independent replication)") was used for + the verdict (OPEN); the ⊘ note was read but did not change the verdict. +- `dismech-causal-replay-v1.md` — has 14 occurrences of ⊘ (the most of any + file in this batch) but they are all inline corrections/retractions of + specific numbers (e.g. "§3b's numbers are SUPERSEDED. They are kept above + (append-only)..."), not a whole-document supersession marker. Status line + "PROPOSED — PLAN/BOARD ONLY" was used directly. +- The 13 non-3DGS files with `status_token = NONE` (listed in Candidate + epiphany #1) were each opened and manually confirmed to have no `Status:` + line anywhere (not just missed by the regex) via a case-insensitive whole- + file `grep -i status` re-check; none contained a differently-formatted + status marker. + +No file was left completely unclassified — every row in the table above has +a verdict. The AMBIGUOUS verdicts (36 total) are 35 "no status line found" +cases plus the 1 `cascade-seal-register-grid-v1.md` case above; none are +unexplained gaps. \ No newline at end of file diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md new file mode 100644 index 000000000..2a6b60c98 --- /dev/null +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md @@ -0,0 +1,580 @@ +# Inventory: `.claude/plans/` — second half (L–Z, entries 104–211 of the sorted directory listing) + +Repo: `/home/user/lance-graph`, checked out on `main` at `aeebfb23`. Range +confirmed: `ls .claude/plans | sort` has 211 entries; entry 104 is +`lance-convergence-staged-migration-v1.md`; this inventory covers entries +104–211 inclusive (108 files, through `wikidata-lazy-spine-hydration-v1.md`). + +Method note (read carefully before trusting any single cell): for every +file the metadata region (first ~80–120 lines) was read with `Read`, the +verbatim `Status:`/`**Status:**` line was extracted (not a substring match — +see the Candidate-epiphanies section for a documented case where naive +substring matching would have misfired), and every `D-`-prefixed identifier +was extracted with a regex, THEN filtered for a documented false-positive +class (see below) before being reported as a `d_id`. The 25 flagged +"special attention" files plus every file with no status line were read in +full or substantially (metadata + body sections + tail). D-id board +cross-checks used `STATUS_BOARD.md`'s own rows, anchored on the row's +leading `| D-id |` cell. + +**D-id extraction false positives found and corrected (read this before +citing any `d_ids` cell as ground truth):** the regex `D-[A-Z][A-Z0-9]*(-[A-Z0-9]+)+` +matches not only real deliverable IDs but also the tail of longer +epiphany-style sentence-IDs that begin `E-...` or `TD-...` — e.g. the +literal text `E-READ-NOT-GREP` contains the substring `D-NOT-GREP`, which +the naive regex reports as if it were a real `D-NOT-GREP` deliverable. +**81 such false positives were found and removed across 43 files** (full +list in Candidate epiphanies). The `d_ids` column below is the +POST-FILTER list. Some remaining entries in that column are still not +literal STATUS_BOARD deliverable rows but this repo's own house style of +citing a full-sentence finding as a `D--N` id (e.g. +`D-A-CONTROL-THAT-CANNOT-LOSE-IS-NO-CONTROL-1`) — these are legitimate +citations in-file, just not STATUS_BOARD rows, so they correctly show as +"not found" in the board cross-check rather than as errors. + +## Table + +| file | title | status_token | version | d_ids | prs | last_commit | verdict | +|---|---|---|---|---|---|---|---| +| `lance-convergence-staged-migration-v1.md` | lance convergence — the STAGED migration plan (v1) | MEASURED / ready-to-execute for stages 0–1; stage 3 is | v1 | D-BSW-0, D-BSW-2, D-LNC-0, D-LNC-1, D-LNC-2, D-LNC-3, D-LNC-4, D-LNC-5, D-LNC-6, D-LNC-7, D-MTS-6 | #1182, #1187, #1189, #1190, #8206*, #8589* | 2026-09-05 | OPEN — active 2026-09-05 arc; D-LNC-0..3 Shipped but D-LNC-4/5/6 + D-LNC-5a/D-MW-P2 still Queued/In PR | +| `lance-graph-business-logic-poc-via-woa-rs-v1.md` | Lance-Graph Business-Logic POC via woa-rs — v1 | Active (Draft) | v1 | D-LGMC-1, D-LGMC-15, D-LGMC-2, D-LGMC-22, D-LGMC-4, D-LGMC-5, D-LGMC-6, D-LGSMB-1, … (+39 more) | #150, #151, #152, #363, #372, #383, #390, #407 | 2026-05-25 | OPEN — Active (Draft) roadmap; references other plans' D-ids, ships nothing itself | +| `lance-graph-in-medcare-rs-v1.md` | lance-graph in MedCare-rs — v1 | Draft | v1 | D-LGMC-1, D-LGMC-10, D-LGMC-11, D-LGMC-12, D-LGMC-13, D-LGMC-14, D-LGMC-15, D-LGMC-16, … (+35 more) | #355 | 2026-05-25 | OPEN — Draft; D-LGMC-* deliverables not confirmed shipped | +| `lance-graph-in-smb-office-rs-v1.md` | lance-graph in smb-office-rs — v1 | Draft | v1 | D-LGMC-4, D-LGSMB-1, D-LGSMB-2, D-LGSMB-3, D-LGSMB-4, D-LGSMB-5, D-LGSMB-6, D-LGSMB-7, D-LGSMB-8, D-LGSMB-9, D-SDR-2, D-UB-2, D-UB-5, D-WLG-3 | #18, #273 | 2026-05-25 | OPEN — Draft; Phase A explicitly "Not started" | +| `lance-graph-in-woa-rs-v1.md` | lance-graph in woa-rs — v1 | Draft | v1 | D-LGMC-1, D-LGMC-2, D-LGMC-21, D-LGMC-4, D-LGSMB-1, D-UB-11, D-UB-4, D-UB-8, D-WLG-1, D-WLG-10..17, D-WLG-2..9 | #275 | 2026-05-25 | OPEN — Draft; 6-phase plan, no shipped confirmation found | +| `lance-graph-ontology-v5.md` | Plan: lance-graph-ontology v5 — post-merge follow-ons | Drafted (2026-05-07). Picks up where v4 left off | v5 | D-ONTO-V5-1 … D-ONTO-V5-15 (15 items) | - | 2026-05-07 | OPEN — 15-deliverable plan; only D-ONTO-V5-9 confirmed shipped (PR #355, per cross-repo citations), rest unconfirmed | +| `lance-graph-rdf-fma-snomed-v1.md` | lance-graph-rdf — FMA / SNOMED CT / RadLex import + named-graph context | plan, not implementation. v1. | v1 | - | #331 | 2026-05-04 | OPEN — explicitly "plan, not implementation" | +| `lance9-datafusion54-upgrade-probe-v1.md` | lance9-datafusion54-upgrade-probe-v1 — what breaks on lance 9 / lancedb 0.33 / DataFusion 54 / Rust 1.97.1 | MEASURED 2026-08-05. Assessment, not a landed migration | v1 | D-THE-NEW-ONE-1 (house-style citation, not a board row) | #14, #19, #93, #179, #243, #244, #245, #351, #891, #895 | 2026-08-05 | CLOSED — the assessment itself is complete per its own status; the actual migration it fed into landed later via `lance-convergence-staged-migration-v1.md` | +| `lf-integration-mapping-v1.md` | LF Integration Mapping — v1 | Active (2026-04-25) | v1 | - | #262, #263, #264 | 2026-04-25 | OPEN — Active, no closure signal found | +| `lite-unified-surrealql-lance-v1.md` | lite-unified-surrealql-lance-v1 — one store + one query surface, behind a feature gate | CONJECTURE / design. "Test via feature gate; do NOT commit" | v1 | - | - | 2026-06-18 | OPEN — CONJECTURE/design, explicit no-commit guardrail | +| `literature-probe-ladder-v1.md` | Literature as falsifier — the probe ladder — v1 | PROPOSED (doc-only). No code, no contract change, no board mutation | v1 | D-LIT-1, D-LIT-2, D-LIT-3, D-LIT-4 (D-LIT-5 in filename only, not a real id) | - | 2026-07-22 | OPEN — PROPOSED; D-LIT-1..4 all Queued on STATUS_BOARD | +| `mailbox-belief-update-and-substrate-test-v1.md` | mailbox-belief-update-and-substrate-test-v1 — "what did I learn" + two-axis substrate test | CONJECTURE / design. 5+3 council COMPLETE. Rides AFTER cycle-aware write | v1 | D-MBX-A3 | - | 2026-06-18 | OPEN — design ratified (council COMPLETE) but code (D-MBX-A3) is Queued on the board | +| `mailbox-cycle-aware-write-contract-v1.md` | mailbox-cycle-aware-write-contract-v1 — every SoA write carries/checks its cycle | CONJECTURE / design. 5+3-gated before code. | v1 | D-MBX-9 | #535 | 2026-06-18 | OPEN — file's own §"Open questions" says ALL 5+3 OQs are resolved, but D-MBX-9 is still Queued on STATUS_BOARD (gates on D-MBX-7/8 + surrealdb + D-PERSONA-5) | +| `mask-algebra-revision-read-v1.md` | mask-algebra-revision-read-v1 — PLAN (DRAFT) | DRAFT, awaiting operator ruling on §5 | v1 | D-ATOM-4, D-MAR-1, D-MAR-2 | #651, #1099 | 2026-08-31 | OPEN — D-MAR-1 Shipped, D-MAR-2 explicitly blocked pending the operator ruling | +| `measure-64k-axes-v1.md` | measure-64k-axes v1 — the corrected five-axis benchmark (operator-specified, 2026-08-05) | NONE | v1 (higher siblings v2/v3/v4 exist) | D-BLW-4, D-KIA-A2 | #445, #879 | 2026-08-05 | SUPERSEDED — v2/v3/v4 siblings exist | +| `measure-64k-axes-v2.md` | measure-64k-axes v2 — rolling epoch closure (operator-specified, 2026-08-05) | NONE | v2 (higher siblings v3/v4 exist) | D-KIA-A2 | #879 | 2026-08-05 | SUPERSEDED — v3/v4 siblings exist | +| `measure-64k-axes-v3.md` | measure-64k-axes v3 — the three arms Stage A0 earned (operator-directed, 2026-08-05) | NONE (read in full — see special note) | v3 (higher sibling v4 exists) | D-KIA-A2 | - | 2026-08-05 | SUPERSEDED — v4 sibling exists; its own M-arm/O-arm results are explicitly folded forward into v4 + a knowledge doc | +| `measure-64k-axes-v4.md` | measure-64k-axes v4 — the hot version window (operator-directed, 2026-08-05) | NONE | v4 (latest) | D-LANCE9-LANCEDB036-REMEASURE, D-SEMANTIC-SLOT (house-style citations) | - | 2026-08-05 | OPEN — latest in the v1-v4 chain, no status line found, no higher sibling | +| `medcare-consumer-pull-thinking-proof-v1.md` | medcare-consumer-pull-thinking-proof-v1 — one real medical thought through the already-live OGAR consumer path | ACTIVE (consolidation + proof target). Date 2026-08-02 | v1 | - | #879 | 2026-08-09 | OPEN — ACTIVE | +| `mul-calibration-not-verdict-v1.md` | Plan: MUL calibrates, it does not adjudicate | PROPOSAL (unbuilt) — 2026-08-26. PLAN/BOARD ONLY, no code | v1 | D-GATE-1, D-GATE-2, D-GATE-3, D-GATE-5, D-MCAL-0, D-MCAL-1..6, D-TSC-1 | #1045, #1052, #1054 | 2026-08-26 | OPEN — D-GATE-1..5 / D-MCAL-0 Queued | +| `mul-consumer-build-gate-v1.md` | Build gate: the D-MCAL arc against its real consumers | GATE RUN — 2026-08-27. Discharges D-MCAL-6 and the second half of… | v1 | D-MCAL-1..6 | #1045, #1065..1069 | 2026-08-27 | OPEN — "GATE RUN" reads as complete, but the file's own §4 states F-MUL-6's second half is "OPEN — not discharged" (MedCare-rs never built against this head) — this is exactly the naive-keyword trap the task warns about | +| `mul-consumer-census-v1.md` | Measurement: per-symbol MUL consumer census | MEASUREMENT COMPLETE — 2026-08-27. Measurement only, no code change | v1 | D-MCAL-1, D-MCAL-6 | #1045, #1052 | 2026-08-27 | CLOSED — measurement-only deliverable explicitly finished; D-MCAL-1/6 read Shipped on the board | +| `mul-ewa-trust-propagation-v1.md` | mul-ewa-trust-propagation-v1 — trust is point-wise today; the sandwich is a CANDIDATE propagation operator | PROPOSED — PLAN/BOARD ONLY. Measure-before-carve | v1 | D-MEP-0, D-MEP-1 | #1065, #1068, #1070, #1072, #1074 | 2026-08-29 | OPEN — D-MEP-0/1 Queued | +| `mul-gate-outcome-vs-ground-v1.md` | Plan: separate the gate OUTCOME from producer-owned GROUND | SUPERSEDED (thesis) 2026-08-26 | v1 | D-GATE-1..6, D-KW-2 | #1045, #1052 | 2026-08-26 | SUPERSEDED — explicit marker in status line | +| `multi-server-cognition-expansion-v1.md` | Multi-Server Cognition — Legacy-Stack Displacement + Expansion Readiness (v1) | PROPOSAL — rationale + expansion-readiness. NOT committed deliverables | v1 | D-CSV-6, D-SDR-26 | - | 2026-05-27 | OPEN | +| `normalized-entity-holy-grail-v1.md` | normalized-entity-holy-grail-v1 — typed unified normalization + Op chain over OGIT/OWL/DOLCE/Odoo | PROPOSAL. The trunk that unifies prior workspace work | v1 | D-NEH-1, D-ODOO-BP-1, D-ODOO-EXT-1 | #411, #420, #426, #427 | 2026-05-28 | OPEN | +| `north-star-integration-v1.md` | North-Star Integration — current state → the two-ViewAngle destination (v1) | RATIFIED (council resolved + gates ratified 2026-06-01) | v1 | D-VIEW-1 | #364, #366, #446, #448, #450 | 2026-06-01 | OPEN — the council DECISION is ratified, but the plan itself says it "enumerates open WIRING DECISIONS" before the actual A3→C7 run; D-VIEW-1 (Shipped) is only one item of the ladder | +| `ocr-canonical-soa-integration-v1.md` | OCR → Canonical SoA Integration v1 | PLANTED 2026-06-15 — design only | v1 | D-OCR-2, D-OCR-50, D-OCR-51, D-OCR-52, D-OCR-53 | #496, #498, #500 | 2026-06-16 | OPEN — D-OCR-50 partially shipped (#498); 51/52/53 and 3 named open decisions (§8) remain | +| `ocr-probes-v1.md` | OCR Transcode — Gating Probes v1 | PLANTED 2026-06-16 — from the 5-specialist framing of #497 | v1 | D-OCR-53 | #459, #496, #497, #498 | 2026-06-16 | OPEN | +| `octopus-causal-cot-audit-v1.md` | Octopus — measurement-first related-work + contract audit | MEASUREMENT REPORT. "No code. No new type. No rename." 2026-08-26 | v1 | D-ACR-7, D-AIF-1, D-AIF-2, D-ECG-1, D-ECG-4, D-ECG-6, D-OCT-1, D-OCT-11 | #1057, #1058 | 2026-08-26 | CLOSED — self-contained finished report by its own explicit claim | +| `odoo-business-logic-blueprint-v1.md` | odoo-business-logic-blueprint-v1 — typed Odoo entity DTOs | PROPOSAL. PREREQUISITE for `odoo-savant-reasoners-v2` Group F | v1 | D-ODOO-BP-1, D-ODOO-SAV-1, D-ODOO-SAV-2 | #407, #411..414, #416, #418..420 | 2026-05-28 | OPEN — the Group F it prerequisites is itself Queued | +| `odoo-classes-bitmask-render-v1.md` | odoo-classes-bitmask-render-v1 — the bounded-weekend fix classes.md prescribes | PLAN (pre-council). Authored 2026-05-30 | v1 | D-ARM-7, D-CHESS-BRINGUP-1, D-CLS-1..9, D-CLS-X | #436, #438, #439 | 2026-05-31 | OPEN — D-ARM-7 cited as "HARD PREREQ" and is still Queued | +| `odoo-savant-reasoners-v1.md` | odoo-savant-reasoners-v1 — lance-graph side of the Odoo richness harvest | PROPOSAL. Picks up the explicit cross-repo handover boundary | v1 (higher sibling v2 exists) | D-ODOO-SAV-1..4 | #412, #413 | 2026-05-27 | SUPERSEDED — v2 sibling exists; v2's own text says "v1 SHIPPED in PR #420" before v2 supersedes it as the active plan | +| `odoo-savant-reasoners-v2.md` | odoo-savant-reasoners-v2 — reshape: `Reasoner` trait → typed composition | PROPOSAL. v1 SHIPPED in PR #420 | v2 | D-ODOO-SAV-1, D-ODOO-SAV-4, D-ODOO-SAV-5 | #411, #414, #416, #418..420 | 2026-05-28 | OPEN — D-ODOO-SAV-5a..5e are all Queued on the board | +| `odoo-savant-roster-v1.md` | odoo-savant-roster-v1 | PROPOSAL (the lance-graph side of the woa-rs Odoo savant delegation) | v1 | D-ODOO-1..5 | #412, #413 | 2026-05-27 | OPEN | +| `odoo-source-extraction-v1.md` | odoo-source-extraction-v1 — TIER-1 Odoo source extraction | SHIPPED (Stage 1 complete 2026-05-28; EXT-1..6 landed; 48/53 entities backed; 5 exemptions documented) | v1 | D-ODOO-BP-1, D-ODOO-EXT, D-ODOO-EXT-1..6 | - | 2026-05-28 | CLOSED — explicit SHIPPED with documented, accounted-for exemptions (not silent incompleteness) | +| `ogar-ar-shape-endgame-v1.md` | ogar-ar-shape-endgame-v1 — increments + gates for the OGAR ontology compiler | PLAN (pre-council); becomes PLAN-RATIFIED after 5+3 verdict | v1 | - (D-ids are Inc-numbers, no board rows minted) | - | 2026-06-19 | OPEN — §11.3 shows all Incs PLAN-RATIFIED except Inc4 (conditional) and Inc5 (F5-real DEFERRED); no PR numbers cite the Incs landing | +| `ogar-sink-in-and-consumer-bridge-removal-v1.md` | Plan v1 — OGAR sink-in + consumer-bridge removal | PROPOSED (doc-only plan). Most of the mechanical migration is already shipped | v1 | D-OVC-1, D-SINK-1..5 | #29, #95..98, #105, #106, #109..111, … (+19 more) | 2026-06-30 | OPEN — underlying bridge-deprecation mechanism is shipped, but this plan's own D-SINK-2/3/5 (finish pull-migration, delete aliases, codebook mint) are not confirmed landed | +| `ogar-vocab-contract-codebook-migration-v1.md` | Migration — OGAR `ogar-vocab` codebook ⇄ `lance-graph-contract` classid (v1) | SHIPPING (2026-06-20). Operator signed off §5; D-OVC-1/2/4 landed | v1 | D-GV2-2, D-OVC-1..5 | #557, #560 | 2026-06-20 | CLOSED — §5 decisions all RESOLVED, D-OVC-1/2/4/5 landed per file body | +| `ogit-cascade-supabase-callcenter-v1.md` | OGIT-Cascade · Supabase Realtime · Callcenter Membrane — v1 | plan, not implementation | v1 | D-CASCADE-V1, D-CASCADE-V1-1..15 | #223, #352 | 2026-05-07 | OPEN | +| `ogit-g-context-bundle-v1.md` | OGIT-G Context Bundle — Tier-1 sub-plan (v1) | NONE (no explicit Status line; APPEND-ONLY governance sub-plan doc) | v1 | D-CASCADE-V1-7, D-OGIT-G-1, D-OGIT-G-2, D-OGIT-G-3, D-ONTO-V5-9 | #29, #98, #355 | 2026-05-12 | AMBIGUOUS — no status line; 3 deliverables described with no shipped/closed marker in the file | +| `open-ideas-fetch-v1.md` | Open-ideas fetch — three stale cards re-derived from the tree (v1) | MEASURED / ready-to-execute — PLANNING ONLY in this PR | v1 | D-OIF-0, D-OIF-1, D-OIF-1-DEC, D-OIF-2, D-OIF-2-DEC, D-OIF-3..7 | #76, #119, #121, #288, #301, #1171, #1185, #1188 | 2026-09-05 | OPEN — mixed board statuses: D-OIF-1 Superseded, D-OIF-1-DEC Withdrawn, most others Queued | +| `oracle-funnel-probe-v1.md` | oracle-funnel-probe-v1 — PROBE-ORACLE-FUNNEL: generate→validate→segment, staged | Stage 0 PRE-REGISTERED 2026-08-05 | v1 | D-BLW-5, D-RLG-1 | #241, #244, #245 | 2026-08-05 | OPEN — D-BLW-5 board status is "PAUSED by operator" (loop half) / payload half partially relaunched; D-RLG-1 "GATED (operator word + API)" | +| `oxigraph-arigraph-cognitive-shader-soa-merge-v1.md` | Oxigraph + AriGraph + Cognitive Shader SoA Merge v1 | "architecture context / implementation prompt … not a claim that the implementation already exists" | v1 | - | - | 2026-05-05 | OPEN — pure architecture prompt, no D-ids, no PRs | +| `palantir-parity-cascade-v2.md` | Palantir Parity · Cascade · DTO Ladder — v2 | plan, not implementation | v2 (no higher sibling) | D-CASCADE-V1-2, D-CASCADE-V1-7, D-PARITY-V2-1..15 | #223, #272, #352 | 2026-05-07 | OPEN | +| `persistence-artifact-backed-commit-v1.md` | persistence-artifact-backed-commit-v1 — the canonical persistence contract | RATIFIED (operator ruling 2026-08-09). Phase A **implemented** on | v1 | - | #911 | 2026-08-09 | OPEN — only Phase A confirmed implemented; status wording implies later phases pending | +| `persistence-cycle-wal-bootstrap-v1.md` | persistence-cycle-wal-bootstrap-v1 — the primitive cycle/WAL seam | ACTIVE (bootstrap SHIPPED in PR #878; upgrade phases PLANNED) | v1 | D-MBX-A6, D-MBX-A6-P1 | #878, #912 | 2026-08-09 | OPEN — bootstrap shipped (D-MBX-A6-P1 Shipped) but upgrade phase D-MBX-A6 is Queued | +| `persistent-nars-kg-v1.md` | Persistent NARS Knowledge Graph — Integration Plan v1 | ACTIVE (2026-07-19) | v1 | D-GRAPH-1, D-INFER-DEDUCTIONS-RELATION-BLIND, D-TRUTH-1 (house-style citations, none found on board — see Mismatches) | #10, #11 | 2026-07-19 | OPEN | +| `pillar11-signature-certification-unification-v1.md` | pillar11-signature-certification-unification-v1 | W0-W4 SHIPPED, W5 DEFERRED (trigger measured, not fired) | v1 | D-SK-A | #289, #291, #350, #1111 | 2026-09-03 | CLOSED — main waves shipped; W5 is an explicit, non-blocking deferral (trigger threshold not yet crossed by real data); follow-up closed 2026-09-03 | +| `polyglot-container-query-membrane-v1.md` | Polyglot Container Query Membrane — SurrealQL AST + DataFusion UDF + Cypher (v1) | RESEARCH MAP + INTEGRATION PLAN. Grounded 2026-06-09 | v1 | D-IDENTITY-1, D-IDENTITY-2, D-MBX-6, D-PG-1..7 | #418, #482, #490 | 2026-06-13 | OPEN — D-PG-1/2/3 + D-MBX-6 Queued on board; a LATER plan (`soa-migration-diff-resolution-2026-06-13.md`) self-flags this one as "superseded in spirit" (author's own words in PR #484), but this file itself carries no such marker | +| `post-438-integration-options-v1.md` | post-438-integration-options-v1 — what to do next (council recalibration input) | OPTIONS LIST (pre-council). Composed 2026-05-30 | v1 | D-ARM-1, D-ARM-13, D-ARM-2, D-ARM-3 | #435..439 | 2026-05-30 | OPEN — D-ARM-1/2/3 Queued, only D-ARM-13 Shipped | +| `post-teardown-buildup-survey-v1.md` | Post-teardown buildup survey v1 — the six families as ingredients | SURVEY, read-only, plan-only (no code, no tenant, no ClassView…) | v1 | D-DCR-1, D-DCR-2, D-POP-1, D-POP-2 | #10, #298, #1134 | 2026-09-03 | OPEN — D-DCR-1/D-POP-1/D-POP-2 Shipped but D-DCR-2 In PR; the survey format is deliberately open-ended | +| `probe-excel-compute-dag-v1.md` | probe-excel-compute-dag-v1 — land `ClassView::compute_dag` on the clean 2-axis grid | CONJECTURE / probe scope. The named first proof for the one Core gap | v1 | - | - | 2026-06-18 | OPEN — not yet run | +| `probe-r2il-live-regfile-v1.md` | PROBE-R2IL-LIVE-REGFILE — plan v1 | GREEN (2026-08-26) — 18/18, seven disable runs red-then-green | v1 | - | - | 2026-08-26 | CLOSED — explicit GREEN, 18/18, disable-run discipline followed | +| `probe-revision-attention-view-1.md` | PROBE-REVISION-ATTENTION-VIEW-1 | NONE (no explicit Status line) | - | D-EDIT-ROUNDTRIP-1 (house-style citation) | #1000 | 2026-08-23 | AMBIGUOUS — thesis/probe doc with no recorded pass/fail outcome in the file | +| `q2-foundry-integration-v1.md` | Q2 Foundry-Equivalent Integration Plan — v1 | Proposed (2026-04-24) | v1 | - | #253..260 | 2026-04-24 | OPEN | +| `r2il-bpe-typed-genetic-recombination-v1.md` | R2IL × BPE as typed genetic recombination over the autopoiesis PALETTES (v1) | PROPOSAL, §7's three falsifiers now RUN | v1 | D-PERSONA-5 | #998 | 2026-08-24 | OPEN — the 3 falsifiers in §7 ran green, but the core recombination mechanism (§3–§5: the 4 operators, contract-checking pass, v3 admission loop) is explicitly still "unbuilt/unprobed" | +| `r2il-machine-semantic-contract-v1.md` | R2IL as the machine-semantic contract — documentation, plan, status, integration | PLAN (v1, 2026-08-25). One measured finding underneath | v1 | - | #285..287, #879, #998, #1012, #1013, #1023 | 2026-08-31 | OPEN — a live, long-running pre-registered probe queue through Q8b; Q2/Q3/Q4/Q5 still unresolved, Q3 is an external OGAR-mint gate | +| `reliability-checklist-arc-v1.md` | reliability-checklist-arc-v1 — integration plan as a LIST OF POSSIBILITIES | PROPOSAL / possibility menu (2026-05-30). NOT a committed sequence | v1 | D-MBX-11, D-MBX-A6, D-MBX-A6-P3 | #433, #439, #445 | 2026-06-14 | OPEN | +| `rosetta-codebook-convergence-v1.md` | rosetta-codebook-convergence v1 — the Bible Rosetta SoA + multi-codebook qualia agreement | PROPOSED (doc-only; D-RCC-1 is the gate, runnable today) | v1 | D-RCC-1..8, D-SCI-1 | #849, #850 | 2026-07-26 | OPEN — D-RCC-2/4/5/6 Queued | +| `rubicon-loco-rung-cognitive-fabric-v1.md` | rubicon-loco-rung-cognitive-fabric-v1 — one clockwork fabric, or a measured NO | PROPOSED — PLAN/BOARD ONLY. SOURCE-FIRST | v1 | D-ACR-1, D-ACR-2, D-ACR-3, D-ACR-7, D-ACR-8, D-R2IL-1, D-R2IL-5, D-RLR-1, D-RLR-2, D-RLR-4, D-RLR-5, D-RLR-6 | #70, #561, #565, #590, #879, #1074..1076, #1112, #1152, #1155, #1157 | 2026-09-03 | OPEN — the plan's own Wave-1 ladder (D-RLR-0..6) is entirely Queued or HELD on STATUS_BOARD; §I's "smallest first Wave" has not run | +| `rung-ladder-grounding-v1.md` | rung-ladder-grounding-v1 — ground agichat's RungShift ladder + CollapseGate as LE-contract | PROPOSAL | v1 | D-RUNG-1..4 | - | 2026-05-26 | OPEN — no board rows found for these ids | +| `rung-mul-grounding-v1.md` | rung-mul-grounding-v1 — the MUL fine-tuned into the ladder as an experience curve | PROPOSAL (follow-on to `rung-ladder-grounding-v1`) | v1 | D-RUNG-MUL-1..5 | - | 2026-05-26 | OPEN | +| `rung-persona-orchestration-v1.md` | rung-persona-orchestration-v1 — time-bound persona orchestration | PROPOSAL (sibling to `rung-mul-grounding-v1`) | v1 | D-PERSONA-1..6, D-RUNG-MUL, D-RUNG-MUL-1 | - | 2026-05-26 | OPEN — D-PERSONA-1 Shipped, 2/3/4/6 Queued, 5 blocked (mixed, mostly open) | +| `scientific-kg-substrate-v1.md` | Scientific-KG substrate — crawl → OCR → terms → reason → MUL (v1, scoping) | PROPOSED — scoping doc, no code | v1 | D-SCI-1..4, D-SCI-INSIGHT, D-SRS-1, D-SRS-3 | - | 2026-07-23 | OPEN — D-SCI-2 Queued, D-SCI-3 Blocked; D-SCI-INSIGHT/D-SRS-1/3 Shipped (mixed) | +| `self-reasoning-substrate-v1.md` | Self-Reasoning Substrate — the graph reasoning about itself (v1) | PROPOSED — doc-only. No code, no contract change, no build surface | v1 | D-SRS-1..4 | - | 2026-07-23 | CLOSED — see Mismatches: the status line is stale, all four D-SRS-* deliverables read Shipped on STATUS_BOARD with real file+test citations | +| `singleton-to-snapshot-nudge-v1.md` | Plan: Singleton → Snapshot Nudge | PROPOSAL | v1 | D-MBX-3, D-MBX-5, D-SNGL-1..7, D-SOA-SNAP-1, D-SOA-SNAP-2, D-SOA-SNAP-5 | - | 2026-06-07 | OPEN — D-SNGL-1/2/4 + D-MBX-3/5 Queued, D-SNGL-3 In progress | +| `soa-32-tenant-awareness-redundancy-v1.md` | Plan — 32-tenant 512-byte SoA (M20) | DRAFT (envelope-auditor gate pending) | v1 | D-MTS-6, D-TRI-1, D-TRI-2 | #717, #722, #725, #727, #729 | 2026-07-21 | OPEN | +| `soa-centroid-attention-field-synthesis-v1.md` | SoA Centroid Attention Field — Unified Synthesis v1 | PLANTED 2026-06-15. Gated on `cycle-coherent-soa-snapshot-v1` | v1 | D-OCR-53 | #495 | 2026-06-16 | OPEN — explicitly gated on a plastic-field COW mechanism that is itself still-authoritative/unimplemented; §6 lists 2 open decisions | +| `soa-migration-diff-resolution-2026-06-13.md` | SoA migration diff resolution — 2026-06-13 | NONE (meta-resolution/audit doc; per-plan status table instead of a top status) | - | D-CSV-1, D-IDENTITY-2, D-MBX-A1, D-MBX-A2, D-SNGL-3 | #383, #384, #386, #434, #470, #477..480, #482, #484, #486..490 | 2026-06-13 | CLOSED — a one-time, self-contained audit snapshot as of its own date; its own deliverable is the resolution itself | +| `soa-value-tenant-migration-v1-harvest.md` | SoA Value-Tenant Migration — Phase-1 Harvest Result | HARVEST (2026-06-25, cont.⁴³). Explicitly "NOT the migration" | - | D-GV2-1 | #500, #610 | 2026-06-25 | CLOSED — the Phase-1 harvest deliverable this doc commissioned is delivered, even though the overall migration continues in v2 | +| `soa-value-tenant-migration-v1.md` | SoA Value-Tenant Migration — Plan v1 (harvest brief + 5+3 sign-off) | BRIEF (2026-06-24). This is NOT the migration | v1 (higher sibling v2 exists) | - | #223, #477, #496, #500, #509, #511, #513, #605, #607 | 2026-06-25 | SUPERSEDED — v2 explicitly "Supersedes the v1 BRIEF's implicit single-pass framing" | +| `soa-value-tenant-migration-v2.md` | SoA Value-Tenant Migration v2 — Operator-Locked Phase Sequencing | OPERATOR-LOCKED SEQUENCING (2026-06-25). Supersedes v1's framing | v2 | D-ENVELOPE-PARSER (house-style citation) | #128, #613 | 2026-06-25 | OPEN — current version; Phase 2 gated on the two 5+3 panels | +| `splat-native-ultrasound-v1.md` | splat-native-ultrasound-v1 — integration plan (debt + future) | PROPOSAL / integration plan. Design-spec only; no code in this plan | v1 | D-MBX-10, D-MBX-A2, D-OSM-2, D-SPLAT-1..14 | #17, #25, #30, #31, #39, #162, #189, #426, #432..434, #467, #470, #7474†, #9663† | 2026-06-05 | OPEN — D-SPLAT-1/10/11 + D-MBX-10/D-OSM-2 all Queued | +| `sprint-5-through-9-roadmap-v1.md` | Sprint 5–9 Roadmap — D-SDR Follow-up to FMA Convergence to Compliance Certification | v1 plan, 2026-05-13 evening | v1 | D-SDR-1, D-SDR-2, D-SDR-3, D-SDR-39, D-SDR-6 | #363 | 2026-05-13 | OPEN — no explicit closure marker found; the plan is old (2026-05-13) relative to the 2026-09-07 session date and likely stale, but that is inferred, not confirmed | +| `sql-spo-ontology-bridge-v1.md` | SQL ↔ SPO Ontology Bridge — v1 Implementation Plan | Active | v1 | - | #308 | 2026-04-30 | OPEN | +| `streaming-arm-nars-discovery-v1.md` | streaming-arm-nars-discovery-v1 — Streaming association-rule discovery → NARS revision | PROPOSAL / integration plan. Spec only; no code in this plan | v1 | D-ARM-1..12, D-MBX-10, D-MBX-7, D-MBX-A1..3, D-ODOO-BP-1, D-ODOO-EXT-2 | #418, #433, #434 | 2026-05-29 | OPEN — D-ARM-1/2/3/10/11/12 all Queued | +| `substrate-comfort-zones-v1.md` | substrate-comfort-zones-v1 — where does each substrate formula feel at home? | RUN, §7 — the pre-registered hypothesis is REFUTED | v1 | D-CZ-0, D-CZ-1, D-CZ-2 (+ D-STRATUM-1, house-style citation) | #926..928, #930, #932, #936, #941, #944, #945, #947 | 2026-08-13 | CLOSED — a completed, definitive measurement (refutation is a result, not an unfinished state) | +| `supabase-subscriber-v1.md` | Supabase-shape Subscriber Flow Wire-up — v1 | In progress (2026-04-24) | v1 | - | - | 2026-04-24 | AMBIGUOUS — very old, generic "In progress" status with no later evidence checked this pass; likely stale but not verifiable without a deeper read | +| `super-domain-rbac-tenancy-v1.md` | Super-Domain RBAC + Multi-Tenancy — v1 | Active | v1 | D-ONTO-V5-5, D-SDR-1, D-SDR-10..15, … (40 total, many not board-tracked) | #275 | 2026-05-13 | OPEN — large multi-phase plan cited widely by other plans; many D-SDR-* rows remain Queued elsewhere on the board | +| `tarski-markov-hhtl-seam-v1.md` | TARSKI-MARKOV-HHTL — open questions register (NOT a plan) | HELD — OPEN QUESTIONS ONLY. Proposes no mechanism, licenses no work | v1 | D-WITNESS-1 (house-style citation) | #1007 | 2026-08-23 | AMBIGUOUS — explicitly "not a plan"; contains one withdrawn proposal plus questions that remain genuinely open | +| `temporal-markov-and-style-classes-v1.md` | temporal-markov-and-style-classes-v1 — the ratified 2026-07-10 cognition arc | ACTIVE (operator-ratified 2026-07-10) | v1 | D-CCF-4, D-EPIPHANY-SIG-1, D-MTS-1..6, D-ORG-1, D-ORG-2, D-SF-FILTER, D-TSC-1..4, D-TTV-1 | #11 | 2026-07-17 | OPEN — see special note; the gating probes D-MTS-1/2/3 are all still Queued on STATUS_BOARD, so the code migration this plan gates has not started | +| `tesseract-rs-ast-dll-codegen-v1.md` | tesseract-rs — AST-DLL C++→Rust Codegen Harness v1 | PLANTED 2026-06-15 v2 — layout IS in scope | v1 | D-OCR-30, D-OCR-31, D-OCR-40, D-OCR-41, D-OCR-42 | #498 | 2026-06-16 | OPEN | +| `tesseract-rs-layout-transcode-v1.md` | tesseract-rs — Layout (textord/ccstruct) 1:1 Transcode v1 | PLANTED 2026-06-15. FAITHFUL 1:1, raw-pointer where C++ is intrusive | v1 | D-OCR-30, D-OCR-31 | - | 2026-06-15 | OPEN | +| `tesseract-rs-recodebeam-transcode-v1.md` | tesseract-rs — recodebeam Decoder 1:1 Transcode v1 | PLANTED 2026-06-15. Decoder transcoded 1:1 over HOSTED posteriors | v1 | D-OCR-16, D-OCR-21, D-OCR-40 | - | 2026-06-15 | OPEN | +| `tesseract-rs-traineddata-gguf-v1.md` | tesseract-rs — traineddata → GGUF → embedanything Host v1 | PLANTED 2026-06-15. The LSTM is HOSTED, not transcoded | v1 | D-OCR-10, D-OCR-15, D-OCR-16, D-OCR-21, D-OCR-40 | - | 2026-06-15 | OPEN | +| `tesseract-rs-transcode-master-v1.md` | Tesseract → tesseract-rs — 1:1 Transcode Master Plan v2 | PLANTED 2026-06-15 v2 — design locked | v1 (filename; body is "v2" of the design) | D-OCR-10, D-OCR-15, D-OCR-16, D-OCR-21, D-OCR-30, D-OCR-31, D-OCR-40, D-OCR-42, D-OCR-50, D-OCR-52, D-OCR-53, D-OCR-NN | #498 | 2026-06-16 | OPEN — design locked but implementation not complete per file (a separate `tesseract-rs` repo carries the real transcode progress, per its own CLAUDE.md, not tracked on this board) | +| `tetrahedral-epiphany-splat-integration-v1.md` | Tetrahedral Epiphany Splat Integration Plan v1 | "integration plan / architecture proposal. No runtime implementation is claimed here" | v1 | - | - | 2026-05-02 | OPEN | +| `thinking-engine-harvest-closure-v1.md` | thinking-engine harvest & closure v1 — harvest every gem, retire the residue, close the chapter | PROPOSAL, plan-only (W0 = this census; no code moves in this PR) | v1 | D-ATOM-1, D-CSV-9, D-HOUSE-4, D-MBX-A6, D-PERSONA-1..6, D-PERT-1, D-REUNIFY-2..6, D-TEH-0..5, D-TRI-1, D-TRI-6, D-TSC-1, D-TSC-3, D-TTV-1, D-V3-W4 | #372, #387, #630, #717, #1051, #1137, #1142..1144, #1151 | 2026-09-03 | OPEN — W0 (this census) is done; the actual per-file fate-assignment / retirement work has not run | +| `thought-cycle-soa-awareness-integration-v1.md` | ThoughtCycleSoA awareness integration plan v1 | "Status: integration plan. No implementation claimed here" | v1 | - | #305, #322, #323 | 2026-05-01 | OPEN | +| `token-value-tenant-v1.md` | token-value-tenant-v1 — a byte-exact span address inside the 40,767-triple stream | PROPOSED — PLAN/BOARD ONLY. Measure-before-carve | v1 | D-TVT-0, D-CONSUMERS-1 (house-style citation) | #10 | 2026-08-28 | OPEN | +| `transcode-extend-core-probe-v1.md` | Transcode EXTEND-CORE + in-env probe slice — v1 (UNDER COUNCIL REVIEW) | PROPOSAL — under 5-consolidate + 3-brutal council review (2026-06-17) | v1 | D-CPP-CODEGEN-1 | #17 | 2026-06-17 | OPEN | +| `triangle-tenants-gestalt-separation-v1.md` | Triangle tenants × separated awareness surfaces × chess quarantine — v1 | DESIGN (operator-directed 2026-07-17; no bytes land with this) | v1 | D-SF-OPPONENT-1, D-SF-OPPONENT-3, D-TRI-1..6, D-TSC-1 | - | 2026-07-17 | OPEN — D-TRI-2/3/4/5 Queued, D-TRI-1 gated, D-TRI-6 In PR | +| `unified-bridge-consumer-migration-v1.md` | Unified-Bridge Consumer Migration — v1 | Draft | v1 | D-ONTO-V5-9, D-SDR-1, D-SDR-2, D-SDR-35, D-UB-1..14 | #355, #407, #408 | 2026-05-25 | OPEN | +| `unified-integration-v1.md` | Plan — Unified Integration: PersonaHub × ONNX × Archetype × MM-CoT × RoleDB | Active — brainstorm phase complete; deliverables defined; no code shipped yet | v1 | - | - | 2026-04-24 | OPEN — explicitly "no code shipped yet" | +| `unified-ogit-architecture-v1.md` | Unified OGIT Architecture — v1 (Master Synthesis) | NONE (master synthesis doc, no single top Status line) | v1 | - | - | 2026-05-12 | OPEN — no top status; §7 "Honest Self-Assessment" lists substantial not-yet-started items itself (no v2 written) | +| `unified-soa-convergence-v1-addendum-2026-05-29-review.md` | unified-soa-convergence-v1 — review addendum (post-merge, 2026-05-29) | REVIEW NOTES — review of the merged plan after PR #434 landed | - | D-MBX-11, D-MBX-9, D-MBX-A1, D-MBX-A2 | #434 | 2026-06-12 | CLOSED — its own §10 in-place edits were applied in the same commit; a self-contained, delivered addendum | +| `unified-soa-convergence-v1.md` | unified-soa-convergence-v1 — THE single little-endian SoA, end-to-end | PROPOSAL / integration plan. Design-spec only; no code in this plan | v1 | D-CE64-MB-1, D-CSV-10, D-CSV-7, D-MBX-2..12, D-MBX-A1..6, D-ODOO-2, D-ODOO-SAV-4, D-PERSONA-5 | #388, #414, #416..418, #433, #434, #445, #477, #490 | 2026-06-14 | OPEN — its own ERRATA note says the §4.2 stack-pin table is stale and points readers to `soa-migration-diff-resolution-2026-06-13.md` for the fuller diff; §1 rulings remain authoritative but substantial content is stale without a top-level SUPERSEDED marker | +| `unified-soa-rubikon-integration-v1.md` | Unified SoA — the Rubikon-model integration (v1) | "Status legend: SHIPPED / PARTIAL / PROPOSED" per-item (no single top status) | v1 | D-MBX-11, D-MBX-12, D-MBX-7, D-MBX-8, D-MBX-9, D-MBX-A6, D-MBX-A6-P3 | #437, #439, #477, #557 | 2026-06-20 | OPEN — a per-item legend, not a closed plan; D-MBX-7/8/9/12/A6 remain Queued | +| `v3-convergence-wiring-v1.md` | V3 Convergence Wiring — Plan v1 (wire, don't invent) | ACTIVE (2026-07-01) | v1 | - | - | 2026-07-02 | OPEN — D1a/D1b/D2/D6 shipped in-session per the file's own D-id table, but D3 Queued, D4 Planned, D7 a "deferred frontier" | +| `v3-substrate-integration-v1.md` | v3-substrate-integration-v1 — pointer stub | POINTER STUB — plan body lives at `.claude/v3/INTEGRATION-PLAN.md` | v1 | - | - | 2026-07-02 | OPEN — redirects to the still-very-active `.claude/v3/` program (see `CLAUDE.md` §V3 SUBSTRATE) | +| `w3-template-mask-v1.md` | W3 — the grammar template as StepMask: no finetuning, three tables — v1 | PROPOSED (doc-only). Gated on D-W3M-1 — no further W3 work | v1 | D-W3M-1, D-W3M-2, D-W3M-3 | - | 2026-07-22 | OPEN — all three Queued | +| `weather-soa-bake-v1.md` | weather-soa-bake-v1 — Zarr → NodeRow, the missing bake | PLAN (2026-08-13). Doc-only. No code shipped, no board file written BY THIS PLAN (own §11 lands its board rows in the same commit) | v1 | D-CZ-0, D-CZ-1, D-CZ-8, D-KIA-A2, D-WXA-1..5, D-WXB-1, D-WXB-4, D-WXC-1, D-WXS-0..12 | #460, #504, #901, #907, #920, #950 | 2026-08-13 | OPEN — its own D-WXS block is mostly Queued, D-WXS-0 explicitly Blocked (operator/OGAR classid mint) | +| `weather-substrate-evaluation-v1.md` | Evaluation Plan — Weather Normalized Substrate: KNOWN vs TO TEST (v1) | ACTIVE (audited 2026-08-11 — §8) | v1 | - | #498, #915, #917, #920..922 | 2026-08-11 | OPEN | +| `weather-substrate-poc-v1.md` | Weather Substrate POC — Plan v1 (encoder bake-off first, forecast second) | PROPOSED (doc-only). No code in this PR | v1 (higher sibling v2 exists) | D-BLW-5, D-HWV-1, D-WX-0..6 | #460, #879, #907 | 2026-08-10 | SUPERSEDED — v2 explicitly "Supersedes `weather-substrate-poc-v1.md`" | +| `weather-substrate-poc-v2.md` | Weather Substrate POC — v2 (jc-gated: representation → hardware → prediction) | PLAN (2026-08-10). Supersedes v1 (PR #914) | v2 (current) | D-WX-0, D-WXA-1..5, D-WXB-1..4, D-WXC-1..5 | #460, #879, #900..902, #907, #913, #914 | 2026-08-11 | OPEN — current version of the weather POC plan | +| `weather-w-probes-v1.md` | weather-w-probes-v1 — the W series as self-contained Sonnet worker briefs | ACTIVE for exploratory probes (W5, W2s-a, W6, W2s-b, W7) | v1 | - | #921, #926, #936, #938, #940 | 2026-08-12 | OPEN | +| `wikidata-lazy-spine-hydration-v1.md` | IMPLEMENTATION PLAN: Wikidata lazy-spine hydration v1 | QUEUED (all D-ids) | v1 | D-ARM-14, D-ARM-7, D-LWS-1..9 | #437, #441, #442, #445 | 2026-06-14 | OPEN — explicitly "nothing shipped yet" | + +`*` = `lance#8206`/`lance#8589` are references to issue/PR numbers in the +**upstream `lance` crate's own GitHub repo**, not this repo's PRs — captured +by the `#\d+` regex without repo context; flagged so the number isn't +misread as a `lance-graph` PR. +`†` = `#7474`/`#9663` in `splat-native-ultrasound-v1.md` read as +out-of-range for this repo's PR numbering (which tops out in the low +1200s elsewhere in this pass) — not individually verified against GitHub +this session; likely also an external/upstream citation or a numeric +artifact, flagged rather than asserted. + +## Special-attention notes + +**`measure-64k-axes-v3.md`** (read in full). This is the third of a four-part +same-day experiment series (v1→v4, all dated 2026-08-05). v3 defines two +arms: **M-arm** (insert a Morton reorder before the seal/WAL write) and +**O-arm** (source the write-side row order from replay instead of the seal). +Both ran. M-arm's own framing requires the verdict to be the *sum* +(`Δtotal = reorder_cost − downstream_savings`), not the reorder cost alone — +the file doesn't give the final number, deferring it to v4. O-arm's result is +a **semantic failure, not just a slow one**: presorting cannot replace the +seal (digest divergence measured), so the standing position becomes "keep +`temporal.rs` as the read-side ordering authority and the seal as the +write-side one, and treat the gap between them as an explicit open research +question." What's left open, concretely: the M-arm net verdict (deferred to +v4), and the specific question "what does the seal compute that +`temporal.rs` does not encode" (four items named, three probes +pre-registered, tracked in a separate knowledge doc, not run in this file). + +**`ocr-canonical-soa-integration-v1.md`.** Design-only sub-plan mapping OCR +tokens onto the canonical `NodeRow`. D-OCR-50 (token→NodeRow mapping) is +partially shipped via `ocr.rs`'s `LayoutBlock::to_node_row` (#498), but three +concrete gaps remain even within D-OCR-50 itself: token-grain nodes don't +exist yet (only block-grain), the HHTL layout-trie is unpopulated (0 depth), +and the OGAR OCR class was never minted (still riding the `0x0000_0000` +bootstrap classid). D-OCR-51 (value-tenant preset) and D-OCR-52 (DeepNSM + +CAM/PQ repair wiring) are unstarted. D-OCR-53 (bit-reproducible golden-file +harness) is explicitly gated on 50+51 landing first. Three open decisions in +§8 (dedicated `ValueTenant::OcrEvidence` vs riding existing tenants; +node-per-token vs node-per-line; character-confusion layer's crate home) are +unresolved. + +**`ogar-ar-shape-endgame-v1.md`.** A 5-increment (Inc1-5) plan promoting five +CONJECTURE rows in a separate doctrine doc to FINDING via probes. Per its own +§11.3 "Final PLAN-RATIFIED status" table: Inc1/2/3a/3b are PLAN-RATIFIED, +Inc4 is PLAN-RATIFIED-conditional (operator must pick weaken-the-gate vs +defer-until-Odoo-arms), and Inc5 is split — the "smoke" half is ratified but +the load-bearing "F5-real" half (one executor pair — NativeLance or +SurrealAst — must do a REAL write, not a stub, with property-fuzzed binding +values) is explicitly DEFERRED. The doctrine's own §10 "litmus" claim stays +CONJECTURE until F5-real runs. No PR numbers appear in the file citing any +Inc as landed, so as read this plan is still pre-execution. + +**`ogar-sink-in-and-consumer-bridge-removal-v1.md`.** Grounds a real, mostly +already-shipped state: all six per-consumer bridges (medcare/odoo/redmine/ +smb/openproject/woa) are already `#[deprecated]` one-line aliases over a +single generic `UnifiedBridge` harness — that half is DONE. What +remains open is layer (b): finishing each consumer's migration off the +deprecated alias onto a direct `Port::class_id()` pull, and only then +deleting the aliases (D-SINK-3), which is explicitly gated on that +consumer-side migration (D-SINK-2) plus porting the `bridge_scope_lock.rs` +cross-namespace-leak test onto the surviving path first. The classid realign +(merged-OSINT/FMA hi/lo ordering) is separately operator-gated and this plan +declines to pre-empt it. + +**`oxigraph-arigraph-cognitive-shader-soa-merge-v1.md`.** Purely an +architecture-context/implementation-prompt document — the file's own second +line says "this is not a claim that the implementation already exists." No +D-ids, no PRs, no board presence. It sketches a 5-PR-shaped roadmap (schema → +witness lanes → search → prefetch → splat stubs) but nothing in the plan +itself, nor anywhere on the board, shows any of those PRs landing. Fully +open, essentially unstarted relative to this specific merge shape. + +**`pillar11-signature-certification-unification-v1.md`.** Unusual in this +set for actually closing out: W0-W4 all measured and shipped, with three of +the plan's OWN pre-registered gates falsified by their own sweeps mid-run and +replaced (documented honestly as "the outcome the probe-first discipline +exists to produce," not hidden). W5 (a memory/time trigger threshold for a +"period-forced" regime) was measured but has **not fired** — the longest +in-tree signature path (4609) is well short of the trigger thresholds +(~11585 / ~33388) — so W5 stays a documented, non-blocking deferral rather +than an open task. A same-day follow-up (Cholesky PSD extension) that had sat +"Queued" for three days after actually shipping was explicitly closed in this +file on 2026-09-03, with a note that a stale Queued row "is a licence for +duplicate work." + +**`probe-r2il-live-regfile-v1.md`.** A genuinely GREEN, closed probe (18/18, +seven disable-runs red-then-green) — but its own §9 makes an important +caveat explicit: **the CI workflow that would run this probe has literally +never executed** (`total_count: 0` on the sibling `r2sleigh` repo's GitHub +Actions) because it targets an unregistered self-hosted runner. Every green +number in the file is a *local* run only. The probe explicitly frames this as +the same "a gate that never runs is not a gate" lesson recorded once already +in `tesseract-rs`. Fixing the CI registration is named as the operator's +call, not done in this plan. + +**`r2il-bpe-typed-genetic-recombination-v1.md`.** §7's three narrow +falsifiers (do splice points exist / does recombination round-trip / does a +recombined candidate produce a distinguishable counterfactual verdict) all +ran and all answered YES with real measured numbers (107/1056 ordered pairs +admit a splice point; 10 distinguishable recombined sequences plus 5 +correctly-silent identity substitutions; a strong recombined splice pair +wins `minority_wins()` against an identical weak majority while an ordinary +baseline correctly does not). But the file is explicit that this proves only +the three narrow questions, not the mechanism: the four recombination +operators (§3), the contract-checking pass (§4), and the v3 +admission/revision loop (§5) remain entirely "unbuilt/unprobed." The +counterfactual leg specifically only exercises the primitive that's +actually shipped (`deposit_counterfactual` + `minority_wins()`); the v3 +`CounterfactualMailbox`/`revise_if_minority_wins` path is still a `todo!()` +stub, blocked on `D-PERSONA-5`. + +**`r2il-machine-semantic-contract-v1.md`.** The largest and most active file +in this half (1427 lines), structured as documentation (§3) + an answer (§4) ++ a long, still-growing, pre-registered probe queue (§6-§11, through "Q8b"). +Open items per its own §8.2 unified queue: **Q2** (a cheap 6502-lift +measurement, settles a conditional) is unresolved; **Q3** (the OGAR mint +decision for a custom memory space) is an EXTERNAL gate blocking "W2"; **Q4** +(op-row carving) "may proceed" but hasn't; **Q5** (the ontology-morphogenesis +probe) has its data half owned by a different repo. The most recent +completed work (Q8, run 2026-08-31, then Q8b) actively falsified the plan's +own hex-topology hypothesis: a degree-1 ablation matched the full +hex-topology arm to four decimal places at 5.5× less memory, meaning "the +six-neighbourness does nothing" — the instrument was reverted and the result +is board-only. Q8b then found the plan's own completion-task metric is not +even expressible on the workspace's real def-use carrier (BPE compresses +92-99.7% of chains to a single symbol, leaving no cued position). So the +plan's own probe apparatus, as specified, needs re-design before the next +question can be asked on it. + +**`rubicon-loco-rung-cognitive-fabric-v1.md`.** A "one clockwork fabric, or +a measured NO" design doc explicitly marked SOURCE-FIRST (every §A-§J claim +carries a file:line). §I names "the ONE smallest first Wave" (`D-RLR-1`): a +non-rung-4 horizon invoking one existing Frozen atom via `ogar_loco`, writing +one alpha row, emitting a typed receipt, and replaying it through +`temporal.rs`. Per STATUS_BOARD, **every D-RLR-0..6 row is Queued or HELD** — +Wave 1 has not run. §J's 11 named falsifiers (F-RLR-1..11) are all still +live gates, not yet exercised. A prior version of this plan's own §F (rung +storage model) had to be corrected mid-session after being written without +reading the mandatory prior plan (`alpha-channel-rung-overlay-v1.md`) — +recorded as a self-caught process failure, not silently fixed. + +**`soa-32-tenant-awareness-redundancy-v1.md`.** DRAFT status, "envelope- +auditor gate pending" — i.e. blocked on a review gate before landing. Its +own §6 is explicit about scope: it does not pre-commit the sibling-lane +count (jc-derived later), does not touch `CausalEdge64`/`EpisodicEdges64` +bit-fields, and does not claim the 64-bit awareness conjecture is wrong — +only that it builds the apparatus to test it. D-TRI-1/D-TRI-2 (the cross- +session convergence points) are Queued/gated on the board. + +**`soa-migration-diff-resolution-2026-06-13.md`.** Not itself a plan — an +audit reconciling nine SoA/BindSpace/identity plans against shipped reality +post-#490. Marked CLOSED here because the audit itself is complete and +dated, but it names several OTHER plans as still genuinely open at the time +of writing: `cognitive-write-roundtrip-substrate-v1.md` ("still-authoritative +… blocked on `SoaEnvelope` impls"), `cycle-coherent-soa-snapshot-v1.md` +("still-authoritative … no implementor yet"), and +`singleton-to-snapshot-nudge-v1.md`'s `D-SNGL-3` ("still queued" as of the +audit — confirmed still "In progress" on the current board, three months +later). It also documents a full retirement list (§3.1): `CollapseGateEmission`, +`MailboxSoA::emit()`, the "Baton" carrier type, and the UUIDv8 `NodeGuid` +wrapper were all planned-then-deleted, useful context for anyone who finds +those names in an OLDER plan and assumes they're still current. + +**`soa-value-tenant-migration-v1.md` / `-v1-harvest.md` / `-v2.md`.** A +three-document chain: v1 is a BRIEF commissioning a harvest; the harvest doc +is the filled inventory (its own deliverable, delivered); v2 is the +operator-locked sequencing that supersedes v1's "implicit single-pass +framing" with a forced two-phase order (identity/key-side V3 migration MUST +precede value-tenant/value-side reshaping, because the envelope parser +resolves `tail_variant` upstream of `value_schema`). v2's own §5 flags that +Phase 1's substance is largely an OGAR/MedCare-rs cross-repo read the harvest +session missed on a casing bug (`/home/user/ogar` vs the real +`/home/user/OGAR`) — so Phase 1's real content is that corrective sweep, not +yet confirmed done in this file. + +**`thinking-engine-harvest-closure-v1.md`.** PROPOSAL, plan-only — its own +scope is explicitly "W0 = this census; no code moves in this PR." The +census itself (§0) is thorough: only two crates depend on +`crates/thinking-engine` in production, and the crate is workspace-excluded +and off CI with ~40 pre-existing clippy lints. Four small process/hygiene +fixes are recorded as already applied within this same census pass (dtype +routing correction, two `pub`→private visibility demotions, a test-count +correction from a stale board citation). But the actual "harvest every gem / +retire the residue" work this plan is named for — assigning a fate to each +of the 51 files — has not started; that's the next session's job. + +**`thought-cycle-soa-awareness-integration-v1.md`.** A short, purely +conceptual integration plan ("No implementation claimed here"). Proposes +splitting the old `Vsa16kF32` carrier into `AwarenessPlane16K` / +`GrammarMarkovLens64` / `ReasoningWitness64` / `ThoughtCycleSoA`, all keyed +on shared `thought_id`/`cycle_id`. No D-ids of its own; entirely design +prose. What's left open is everything — this is a naming/shape proposal, not +a build plan with a status ladder. + +**`unified-soa-convergence-v1.md`.** The 2026-05-29 foundational SoA plan +("five layered rulings"). Its own top-of-file ERRATA (added post-#490) is +the load-bearing fact: §1's rulings are still cited as authoritative +(anchor for `E-SOA-IS-THE-ONLY`), but §4.2's stack-pin table is stale (lance +bumped past what's written here at least twice since), and the fuller +picture lives in `soa-migration-diff-resolution-2026-06-13.md`. This plan is +therefore best read as "doctrine current, mechanics stale" rather than +either fully open or fully superseded — no single top-level marker settles +it either way, which is itself worth flagging for anyone citing this file's +stack-pin table as current. + +**`unified-soa-rubikon-integration-v1.md`.** Uses an explicit per-item status +legend (✅ SHIPPED / ◐ PARTIAL / ☐ PROPOSED) rather than one top status. §6 +"Honest status (no overclaim)" cross-references itself against +`unified-soa-convergence-v1.md`'s D-MBX ladder — most of those (D-MBX-7/8/9/ +12/A6) remain Queued on the current board. The plan's real headline claim +(SurrealQL as "a co-equal lens on the one SoA, gated only by OQ-11.6") is +gated on uncommenting a fork dependency and filling in one stub function +(`read_via_kv_lance`) — a small, specific, still-open item. + +**`v3-convergence-wiring-v1.md`.** ACTIVE, "wire, don't invent" — the +organizing finding is that most V3 substrate gaps are unwired seams, not +missing machinery. Of its 7 named deliverables, D1a/D1b/D2/D6 are marked +shipped in the same session that wrote the plan; D3 (q2 render probe) is +push-gated on a separate repo; D4 (registry consolidation) is merely +planned; D5 is recorded as an open issue (`ISS-Q2-CASCADE3-NIBBLE-ANCESTRY`) +rather than resolved; D7 (rig/rs-graph-llm orchestration loop) is explicitly +named "the only fully unwired angle" and deferred until the operator opens +it — that's the one item genuinely wide open. + +**`weather-soa-bake-v1.md`.** Opens by correcting the framing it was +commissioned under: the predecessor plan (`weather-substrate-poc-v2.md`) +named a `D-WXA-5` gate and D-ids that were **never minted onto +STATUS_BOARD.md at all** (`grep -c "WXA"` = 0, independently re-verified by +the orchestrator) — a concrete, named instance of the +`ISS-PLAN-TRACKING-IS-UNENFORCED` board-hygiene gap this repo already +tracks. It further argues that gate's own threshold (Spearman ρ ≥ 0.98) is +"at serious risk of being VACUOUS" because a sibling plan already measured +that ρ *saturates* near 1.0 on smooth pressure fields regardless of +representation quality. This plan's own D-WXS block is Queued end-to-end +except D-WXS-0 (classid mint), which is explicitly Blocked on +operator/OGAR — so nothing in the actual bake has landed yet. + +## Counts + +| Verdict | Count | +|---|---| +| OPEN | 85 | +| CLOSED | 12 | +| SUPERSEDED | 7 | +| AMBIGUOUS | 4 | +| **Total** | **108** | + +CLOSED: `lance9-datafusion54-upgrade-probe-v1.md`, `mul-consumer-census-v1.md`, +`octopus-causal-cot-audit-v1.md`, `odoo-source-extraction-v1.md`, +`ogar-vocab-contract-codebook-migration-v1.md`, +`pillar11-signature-certification-unification-v1.md`, +`probe-r2il-live-regfile-v1.md`, `self-reasoning-substrate-v1.md`, +`soa-migration-diff-resolution-2026-06-13.md`, +`soa-value-tenant-migration-v1-harvest.md`, `substrate-comfort-zones-v1.md`, +`unified-soa-convergence-v1-addendum-2026-05-29-review.md`. + +SUPERSEDED: `measure-64k-axes-v1.md`, `measure-64k-axes-v2.md`, +`measure-64k-axes-v3.md`, `mul-gate-outcome-vs-ground-v1.md`, +`odoo-savant-reasoners-v1.md`, `soa-value-tenant-migration-v1.md`, +`weather-substrate-poc-v1.md`. + +AMBIGUOUS: `ogit-g-context-bundle-v1.md`, `probe-revision-attention-view-1.md`, +`supabase-subscriber-v1.md`, `tarski-markov-hhtl-seam-v1.md`. + +## Mismatches + +Two genuine plan-vs-board disagreements found (a much smaller set than the +raw automated pass produced — see Candidate epiphanies for why most +apparent "mismatches" were discarded as false positives from naive keyword +matching): + +1. **`self-reasoning-substrate-v1.md`** — the file's own status line reads + "PROPOSED — doc-only. No code, no contract change, no build surface" + (2026-07-23). STATUS_BOARD.md shows all four of its own named + deliverables as Shipped, each with a real source file and test count: + `D-SRS-1` (Shipped, `src/reason.rs` + 7 tests), `D-SRS-2` (Shipped, + `src/{shape,ancestry}.rs` + 63 tests), `D-SRS-3` (Shipped — falsifier + fired, conjecture not confirmed), `D-SRS-4` (Shipped — CONFIRMED + positive, `src/introspect.rs` + 77 tests). The plan's own header text is + stale relative to what actually shipped under its own D-ids. + +2. **`persistent-nars-kg-v1.md`** and **`self-reasoning-substrate-v1.md`** + both cite `D-GRAPH-1`, `D-TRUTH-1`, and `D-INFER-DEDUCTIONS-RELATION-BLIND` + as deliverables — none of the three appears anywhere in + `STATUS_BOARD.md` (confirmed by direct grep, not just the automated + D-id map). Either these were never minted as board rows, or they were + later renamed/absorbed into a different D-id family without a pointer + left behind in either plan. + +No other file in this range showed a plan claiming a specific D-id is +shipped/complete while STATUS_BOARD.md shows that same D-id Queued/Blocked/ +Withdrawn — every apparent case surfaced by the first automated pass +resolved, on inspection, to either (a) the plan's own wording already +matching the board (e.g. `persistence-cycle-wal-bootstrap-v1.md` says +"bootstrap SHIPPED … upgrade phases PLANNED," and the board shows exactly +that split), or (b) a keyword false-positive in the crude classifier (see +Candidate epiphanies #1). + +## Candidate epiphanies + +Facts observed this pass, with counts and filenames, no speculation: + +1. **Naive substring/keyword matching on "SHIPPED" misfires repeatedly, + confirming the exact risk this repo's own `CLAUDE.md` documents for + `ARCHIVE?` routing.** A crude classifier that flags a plan CLOSED + whenever its status text contains "SHIPPED" produced false positives on + at least: `weather-soa-bake-v1.md` (status literally says "**No** code + shipped"), `unified-integration-v1.md` ("**no** code shipped yet"), + `unified-soa-rubikon-integration-v1.md` (the word appears only inside a + ✅/◐/☐ status-legend definition, not as an overall claim), and + `mul-consumer-build-gate-v1.md` (status says "GATE RUN," which reads as + complete, but the file's own §4 explicitly labels one falsifier "OPEN — + not discharged"). Every verdict in the table above was therefore set + from a full read of the status line in context, not a keyword scan. + +2. **The `D-[A-Z]+(-[A-Z0-9]+)+` extraction regex produces real false + positives by matching the tail of a longer `E-`/`TD-`-prefixed + sentence-id.** Confirmed 81 false-positive "D-ids" across 43 files this + pass (e.g. `E-READ-NOT-GREP` yields a spurious `D-NOT-GREP` in + `north-star-integration-v1.md`; `E-A-CONTROL-THAT-CANNOT-LOSE-IS-NO- + CONTROL-1` yields a spurious `D-A-CONTROL-THAT-CANNOT-LOSE-IS-NO- + CONTROL-1` in `mul-consumer-build-gate-v1.md`). All were detected by + checking whether the character immediately preceding the match is + alphabetic, and filtered out of the `d_ids` column above. Files most + affected: `octopus-causal-cot-audit-v1.md` (4), `r2il-machine-semantic- + contract-v1.md` (8), `thinking-engine-harvest-closure-v1.md` (5). + +3. **This repo's own house style also uses a second, legitimate pattern + that LOOKS like the same false-positive: `D--N` as a + real in-plan citation for a finding**, distinct from a STATUS_BOARD + deliverable row (e.g. `D-IS-BOUND-TO-ITS-STATISTIC-AND-ITS-SAMPLE-1` in + `pillar11-signature-certification-unification-v1.md`, which does + correspond to a real, later-cited board finding). These pass the + word-boundary filter (nothing alphabetic precedes them) but still show + as "not found" in a literal STATUS_BOARD row lookup, because they're + findings citations, not deliverable ids. A future automated pass should + not treat "not found on STATUS_BOARD" as itself evidence of a tracking + gap without checking whether the id is grammatically a deliverable or a + sentence-citation. + +4. **Some `#NNNN` citations reference an upstream repo's issue/PR tracker, + not this repo's own.** `lance-convergence-staged-migration-v1.md` cites + `lance#8206` and `lance#8589` (the upstream `lance` crate's GitHub + issues) using bare `#NNNN` prose elsewhere in the same paragraph — a + naive `#\d+` PR-citation regex captures these indistinguishably from + this repo's own PR numbers. Flagged in the table rather than silently + conflated. + +5. **Six explicit `-v1`→`-v2`(→`-v3`→`-v4`) supersession chains exist in + this half**, and in every case the higher-numbered file states the + supersession explicitly in its own status line (never silent): the + `measure-64k-axes` v1→v2→v3→v4 chain (3 files marked SUPERSEDED, 1 + current), `odoo-savant-reasoners` v1→v2, `soa-value-tenant-migration` + v1→v2, and `weather-substrate-poc` v1→v2. No case of a stale + lower-version file *without* a superseding sibling's explicit marker + was found in this half (contrast with `polyglot-container-query- + membrane-v1.md` below, which has no version sibling but is + self-flagged as superseded by a THIRD file). + +6. **A "superseded in spirit" self-flag lives in the superseding author's + PR body, not in the superseded plan file itself, and not reachable by + filename/version pattern.** `polyglot-container-query-membrane-v1.md` + carries no `-v2` sibling and no `SUPERSEDED` marker of its own, but a + LATER, unrelated-filename plan (`soa-migration-diff-resolution- + 2026-06-13.md`, itself in this half) quotes the author of PR #484 + self-flagging it as "superseded in discussion by the self-describing- + key convergence." This repo's supersession tracking is therefore not + fully captured by version-suffix conventions alone — some supersessions + only exist as prose in a third document. + +7. **15 files in this half (of 108) have no verbatim `Status:`/`**Status:**` + line at all**, requiring a fallback read of the opening paragraph or + (for 2 files) a full read: `ogit-g-context-bundle-v1.md`, + `probe-revision-attention-view-1.md`, `soa-migration-diff-resolution- + 2026-06-13.md`, `tarski-markov-hhtl-seam-v1.md` (has a bare `Status:` + without bold markers, on its own line), `token-value-tenant-v1.md` + (same bare-Status pattern), `unified-ogit-architecture-v1.md`, + `unified-soa-rubikon-integration-v1.md` (has a "Status legend" instead + of a single status), `v3-substrate-integration-v1.md`, `wikidata-lazy- + spine-hydration-v1.md` (bare `Status:` pattern), plus the four + `measure-64k-axes-v{1,2,3,4}.md` files (all four have zero Status line + of any form), `mul-ewa-trust-propagation-v1.md`, + `r2il-bpe-typed-genetic-recombination-v1.md`, and + `soa-value-tenant-migration-v1-harvest.md` (all three of the last group + use a bare `Status:` — no `**` bold — that a `\*\*Status:?\*\*` regex + alone would also miss). + +8. **`weather-soa-bake-v1.md` documents, as its own §0.1, a real historical + instance of exactly the D-id/board-tracking gap this task's inventory + method is designed to catch**: it names a predecessor plan + (`weather-substrate-poc-v2.md`) whose entire `D-WXA-*`/`D-WXB-*`/ + `D-WXC-*` deliverable ladder was never mirrored onto `STATUS_BOARD.md`, + and states this as the plausible mechanism by which "a whole arc ran + past [a gate] in Python without anyone tripping over an unmet gate." + This is a first-party confirmation, from inside the corpus, that + plan-D-ids not appearing on the board is a known, recurring failure + mode here — not just an artifact of this inventory's own extraction + method. + +## Unread/uncertain + +- **`sprint-5-through-9-roadmap-v1.md`** — only the metadata region was + read (status line + head). It is dated 2026-05-13, four months before the + current session date (2026-09-07), and cites only D-SDR-* ids that are + extensively referenced (and partially shipped) across many other, later + plans in this corpus, but this specific file's own closure state was not + independently verified against a full read. Classified OPEN rather than + CLOSED/SUPERSEDED purely because no explicit marker was found, not + because staleness was ruled out. +- **`supabase-subscriber-v1.md`** — very short metadata read only (status + line: "In progress (2026-04-24)", the oldest file in this half along with + a few April 2026 plans). No later citation of this plan was found + elsewhere in the corpus during this pass, which is itself mildly + suggestive of abandonment, but that absence-of-citation check was not + exhaustive (no full-corpus grep run specifically for this filename). + Marked AMBIGUOUS rather than a guessed CLOSED or SUPERSEDED. +- **`tesseract-rs-*` five files** (`-ast-dll-codegen-v1.md`, + `-layout-transcode-v1.md`, `-recodebeam-transcode-v1.md`, + `-traineddata-gguf-v1.md`, `-transcode-master-v1.md`) — all read at + metadata-region depth only (title/status/D-ids), not in full, because the + actual transcode progress they describe is tracked in a SEPARATE sibling + repo (`tesseract-rs`, which has its own extensive `CLAUDE.md` recording + many shipped leaves — Leaf 1 through the full image-to-text pipeline — + entirely outside this board's D-id namespace). All five are marked OPEN + here on the lance-graph side purely because no D-OCR-* row on THIS + repo's STATUS_BOARD.md confirms closure; the real, much more advanced + status lives in the other repo and was not cross-checked this pass + (out of scope for a `.claude/plans/` inventory of this repo). +- **`unified-ogit-architecture-v1.md`** — the 607-line master-synthesis + file's §0-§4 and §7 (self-assessment) were read; §5 ("Proof of Vision") + and §6 ("Cross-References") were not read in full. The OPEN verdict rests + on §7's own admission of open items, which should hold regardless, but a + full read of §5/§6 was not performed. \ No newline at end of file diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md new file mode 100644 index 000000000..6e09885e3 --- /dev/null +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md @@ -0,0 +1,648 @@ +# Inventory: top-level `.claude/*.md` + board dashboards (lance-graph) + +Repo: `/home/user/lance-graph`, checked out on `main` at `aeebfb23`. All counts below are +mechanically derived (git log / grep / python parse) except where marked otherwise. + +--- + +## Top-level docs table + +71 files at `.claude/*.md` (top level only). Columns: filename | last commit date | role +(one line, from head-of-file) | self-referential marker (⊘/SUPERSEDED/DEPRECATED/RETIRED/ +historical, applied to the FILE ITSELF — not incidental text matches) | cross-ref count +(files under `.claude/` that mention this filename, excluding itself). + +| File | Last commit | Role | Marker | Refs | +|---|---|---|---|---| +| AGENT_COORDINATION.md | 2026-04-25 | Governance doc: the 3 coordination layers agents use (git log / PR descriptions / EPIPHANIES.md / CROSS_SESSION_BROADCAST.md) | — | 1 | +| AGI_DESIGN.md | 2026-04-07 | AGI design note: 4D×16Kbit cluster resonance + neural meta-learning (ONNX micro-learner correcting u8 lookup) | — | 1 | +| ARCHITECTURE_THOUGHT_ENGINE.md | 2026-04-18 | Architecture: "16M RISC Thought Engine" (MatVec/cycle, 3-layer 64/256/4096 branching) | — | 2 | +| ATTENTION_MASK_AUDIT_2026_08_21.md | 2026-08-21 | D-ACR-0 audit report: `attention_mask.rs` graded EXISTS-UNCALLED for the eye-tracking use case | — | 6 | +| BELICHTUNGSMESSER.md | 2026-03-14 | Explainer: "HDR Popcount-Stacking Early-Exit Distance Cascade" (exposure-meter early-exit search) | — | 3 | +| BF16_SEMIRING_EPIPHANIES.md | 2026-03-15 | Research-report epiphanies: BF16 + semirings + binary planes (5:2 bitwise/float split) | — | 4 | +| BGZ17_ELEVEN_SEVENTEEN_RATIONALE.md | 2026-04-11 | Research note: why bgz17 uses 11/17 golden-step constants, read from source | — | 2 | +| BOOT.md | 2026-07-22 | THE session entry point / mandatory read-order doc (canonical bootstrap, named in root CLAUDE.md) | — | 62 | +| CALIBRATION_REPORT_2026_04_03.md | 2026-04-03 | Calibration report: bgz17→f32 bridge + real-data validation (OpenChat/Llama4/Jina v3) | — | 0 | +| CALIBRATION_STATUS_GROUND_TRUTH.md | 2026-04-06 | "Ground truth" correction doc, meant to be read before SESSION_*.md prompts (**named in root CLAUDE.md Knowledge Base**) | — | 5 | +| CODING_PRACTICES.md | 2026-04-21 | Coding-pattern checklist from EmbedAnything (**named in root CLAUDE.md Knowledge Base**) | — | 3 | +| CROSS_REPO_AUDIT_2026_04_01.md | 2026-04-18 | Cross-repo audit (lance-graph/ndarray/q2/AriGraph), 8 parallel agents | (false-positive: "DEPRECATED" is a legend-table cell, not a self-marker) | 3 | +| DECISION_SPO_ARIGRAPH.md | 2026-05-07 | Binding decision doc: SPO-1 crate composition, Option B federated two-layer cache | — | 8 | +| DEEPNSM_CAM_REFERENCE.md | 2026-03-25 | Quick reference for DeepNSM-CAM semantic-lookup layer (data file locations) | — | 3 | +| DEEP_ADJACENT_EXPLORATION.md | 2026-03-15 | Notes on dropped algorithms / RISC design / adjacent research (RDF-3X etc.) | — | 3 | +| DEVELOPMENT_STAGES.md | 2026-04-06 | RISC Thought Engine dev-stage tracker with empirical results | — | 1 | +| DISTANCE_METRIC_INVENTORY.md | 2026-04-18 | Table: valid distance metric per data type across the stack | — | 3 | +| FALKORDB_ANALYSIS.md | 2026-03-16 | Competitive analysis of FalkorDB informing lance-graph design | — | 3 | +| FINAL_STACK.md | 2026-03-22 | Session summary naming the "final" 3-repo stack (ndarray/lance-graph/rs-graph-llm) | — | 3 | +| FIX_BLASGRAPH_SPO.md | 2026-03-13 | Proposal: unify BlasGraph BitVec + SPO Fingerprint into one type | — | 3 | +| GPU_CPU_SPLIT_ARCHITECTURE.md | 2026-03-15 | Architectural epiphany: GPU tensor cores for NARS truth revision, not cosine search | — | 3 | +| HANDOVER_CALIBRATION_SESSION.md | 2026-04-05 | Session handover: calibration/encoding/Cronbach-alpha validation hypotheses | — | 1 | +| HANDOVER_MAVERICK_SESSION.md | 2026-04-05 | Session handover: Llama4 Maverick 128-expert MoE + temperature fix | — | 1 | +| HANDOVER_NEXT_SESSION.md | 2026-04-06 | Handover: forward-pass branch graph + vision + LoRA (**named in root CLAUDE.md Prior art**) | — | 0 (see note) | +| HANDOVER_SIGNED_SESSION.md | 2026-04-06 | Session-stats handover for the "signed session" (17 PRs, 294 tests) | — | 1 | +| IDEA_JOURNAL_2026_04_29_FUTURE_PILLARS.md | 2026-04-29 | Idea journal (pillars 7/8/9); ideas re-filed to canonical board | **historical** ("preserved for historical reference") | 4 | +| IDEA_JOURNAL_2026_04_29_STREAMING_HYDRATION.md | 2026-04-29 | Idea journal (streaming-hydration/fractal codec); ideas re-filed to canonical board | **historical** ("preserved for historical reference") | 4 | +| INTEGRATIONSPLAN_2026_04_01.md | 2026-04-18 | Integration plan: "quick wins to AGI stack" (**named in root CLAUDE.md Prior art**) | — | 2 | +| INTEGRATION_SESSIONS.md | 2026-03-15 | Session-prompt index, inventory→wired-system (**named in root CLAUDE.md Prior art**) | — | 3 | +| INVARIANT_MATRIX_RESEARCH.md | 2026-04-11 | Research note: "which invariant does each lens preserve" reframing | — | 4 | +| INVENTORY_MAP.md | 2026-03-15 | Complete inventory of rustynum-core + lance-graph substrate (**named in root CLAUDE.md Prior art**) | — | 4 | +| KNOWLEDGE_SYNC_SIGNED_SESSION.md | 2026-04-05 | Knowledge-sync doc: "33% error" SiLU-gate correction finding (**named in root CLAUDE.md Prior art**) | — | 0 (see note) | +| LANE_AKKUMULATOR.md | 2026-04-07 | L0-L4 multi-lane lookup-prediction accumulator | — | 1 | +| LANGGRAPH_CRATE_STRUCTURE.md | 2026-03-16 | Python LangGraph package layout → recommended Rust crate structure | — | 0 | +| LANGGRAPH_FULL_INVENTORY.md | 2026-03-16 | Full Python LangGraph → Rust API mapping table | — | 0 | +| LANGGRAPH_OUR_ADDITIONS.md | 2026-03-16 | Rust graph-flow features beyond Python LangGraph's set | (false-positive: "historical state" is prose describing a feature) | 0 | +| LANGGRAPH_PARITY_CHECKLIST.md | 2026-03-16 | DONE/PARTIAL/MISSING parity checklist vs Python LangGraph | — | 0 | +| LANGGRAPH_TRANSCODING_MAP.md | 2026-03-16 | Python→Rust type/function transcoding map for LangGraph | — | 0 | +| ONE_FORTIETH_SIGMA_LENS.md | 2026-04-11 | Research note: "1/40 σ lens" as invariant-matrix column | — | 3 | +| OVERLOOKED_THREADS.md | 2026-03-15 | Catalog of glossed-over research threads (tropical attention etc.) | — | 2 | +| PLAN_BF16_DISTANCE_TABLES.md | 2026-04-06 | Plan: replace u8/i8 distance tables with BF16 tables (**named in root CLAUDE.md Knowledge Base**) | — | 1 | +| RECON_ONTOLOGY_CRATE.md | 2026-05-07 | Phase-1 recon: verifying "already shipped" claims before ontology-crate work | — | 6 | +| RESEARCH_REFERENCE.md | 2026-03-15 | Prior-art catalog ("what to steal"), e.g. DreamerV3 binary RL | — | 2 | +| RESEARCH_THREADS.md | 2026-03-15 | Actionable research threads/connections (DreamerV3 proves binary is better) | — | 4 | +| RING_PERTURBATION_PROPAGATION.md | 2026-04-11 | Research note: ring perturbation awareness propagation (angular half of 1/40σ) | — | 3 | +| RISC_THOUGHT_ENGINE_AGI_ROADMAP.md | 2026-04-06 | AGI roadmap + ground-truth results for the RISC Thought Engine | — | 2 | +| SCOPE_E_FINDINGS.md | 2026-03-23 | API-surface inventory for wiring lance-graph+ndarray+rs-graph-llm | — | 2 | +| SESSION_2026_03_25_CROSS_REFERENCE.md | 2026-03-25 | Cross-reference summary of a big session (thinking-styles-as-graph, DeepNSM, CAM-PQ) | — | 1 | +| SESSION_B_HDR_RENAME.md | 2026-03-22 | Completed session prompt: rename LightMeter→hdr::Cascade (title: "✅ DONE") | done/complete (not ⊘/SUPERSEDED-family) | 2 | +| SESSION_D_LENS_CORRECTION.md | 2026-03-14 | Session prompt: gamma+cushion lens correction + boundary fold | — | 2 | +| SESSION_FALKORDB_CROSSCHECK.md | 2026-03-22 | Session prompt superseded by a `.claude/prompts/` file, kept for reference | **SUPERSEDED** (in title) | 2 | +| SESSION_J_PACKED_DATABASE.md | 2026-03-15 | Session prompt: PackedDatabase panel-packing for cascade search | — | 2 | +| SESSION_LANCE_ECOSYSTEM_INVENTORY.md | 2026-03-15 | Session prompt: inventory unused original-Lance-ecosystem capability | — | 3 | +| SESSION_LANGGRAPH_ORCHESTRATION.md | 2026-03-15 | Session prompt: replace Layer-4 chaos with LangGraph-style execution model | — | 2 | +| STATUSMATRIX.md | 2026-04-07 | RISC Thought Engine status matrix (German), empirically-proven results | — | 1 | +| TECHNICAL_DEBT_SIGNED_SESSION.md | 2026-04-06 | Tech-debt review of the signed session (**named in root CLAUDE.md Knowledge Base**) | — | 2 | +| UNIFIED_HDR_RENAME_AND_CROSSPOLINATE.md | 2026-03-14 | Combined rename + cross-pollination plan (rustynum ↔ lance-graph) | (false-positive: "Deprecated wrapper" describes a code fn, not the doc) | 2 | +| VISION_ORCHESTRATED_THINKING.md | 2026-03-15 | Vision: zero-copy orchestrated thinking (LangGraph+Lance+binary planes) | — | 3 | +| WIKIDATA_EXTRACTION_PLAN.md | 2026-04-06 | Plan: streaming Wikidata→AriGraph hydration via SPARQL | — | 0 | +| agent2agent-orchestrator-prompt.md | 2026-04-11 | Orchestrator prompt for a small agent ensemble (ripple-architecture doctrine) | — | 3 | +| blackboard-ripple-architecture-changelog.md | 2026-04-18 | Changelog for the "Ripple Architecture" exploration | — | 5 | +| blackboard.md | 2026-03-23 | "Polyglot Notebook" single-binary architecture doc (marimo/graph-notebook transcode plan) | — | 12 | +| foundry-roadmap.md | 2026-04-29 | DRAFT unified SMB-Office+MedCare Foundry parity roadmap | not-yet-historical (plans to become HISTORICAL after PR-5; currently still DRAFT) | 6 | +| knowledge.md | 2026-04-18 | "Knowledge Spine" doctrine doc for the ripple-architecture cognition-stack framing | — | 6 | +| medcare-foundry-vision.md | 2026-04-30 | Client-facing architectural-vision draft for MedCare Foundry | — | 4 | +| pattern.md | 2026-05-06 | APPEND-ONLY SoA/DTO graph-traversal primer, "read by every session" touching SoA/DTO | — | 17 | +| patterns.md | 2026-05-12 | "READ FIRST" SoA/DTO navigation guide, near-duplicate topic of `pattern.md` | — | 30 | +| probe_m1_result_2026_04_11.md | 2026-04-11 | Probe M1 result: FAIL with k=4 surprise (clustering-method panel) | — | 0 | +| prompt-for-other-session-additive.md | 2026-04-18 | Cross-session prompt: work additively on ripple architecture | — | 0 | +| ripple-file-index.md | 2026-04-18 | File index for a parallel session entering the ripple-architecture work | — | 1 | +| session_2026_04_11_bf16_hhtl_combined_research.md | 2026-04-11 | Non-canonical research index integrating 4 session notes (BF16-HHTL thread) | — | 1 | + +Note on the 0-ref "named in CLAUDE.md" rows: `HANDOVER_NEXT_SESSION.md` and +`KNOWLEDGE_SYNC_SIGNED_SESSION.md` have zero cross-references *inside* `.claude/*.md` +files, but the root `CLAUDE.md`'s "Prior art" list names both explicitly by filename — +so they are referenced, just from outside the `.claude/` scope the grep covered. + +--- + +## Orphans + +ORPHAN = zero `.claude/`-internal cross-references AND no self-referential marker AND +not named in root `CLAUDE.md`'s "Prior art" / "Knowledge Base" lists (9 files named there; +`SESSION_CAPSTONE.md` is also named but does not exist in this repo — a stale citation in +CLAUDE.md itself, worth flagging). + +**9 true orphans:** + +1. `CALIBRATION_REPORT_2026_04_03.md` +2. `LANGGRAPH_CRATE_STRUCTURE.md` +3. `LANGGRAPH_FULL_INVENTORY.md` +4. `LANGGRAPH_OUR_ADDITIONS.md` +5. `LANGGRAPH_PARITY_CHECKLIST.md` +6. `LANGGRAPH_TRANSCODING_MAP.md` +7. `WIKIDATA_EXTRACTION_PLAN.md` +8. `probe_m1_result_2026_04_11.md` +9. `prompt-for-other-session-additive.md` + +The 5-file `LANGGRAPH_*` cluster is a coherent orphaned unit — a complete Python→Rust +LangGraph parity/mapping doc-set with zero inbound references from any other `.claude/` +doc, and CLAUDE.md's "Prior art" list does not name any of them. + +--- + +## Stale candidates + +STALE-CANDIDATE = last commit before 2026-06-01 AND not in root CLAUDE.md's Prior-art/ +Knowledge-Base lists. Root CLAUDE.md itself says "61 top-level docs" in its Prior-art +section — the real count is **71**, so that count in CLAUDE.md is itself stale (10 files +added since it was last touched). + +**61 files** meet the stale-candidate bar (every top-level doc except `BOOT.md`, +`ATTENTION_MASK_AUDIT_2026_08_21.md`, and the 9 files named in CLAUDE.md's lists — all +69 non-recent files minus those 8 existing named exemptions). Effectively: **every +top-level doc in this directory except BOOT.md and the 2026-08-21 audit report is >3 +months old**, and the vast majority (61/71 = 86%) predate June 2026 entirely. The +directory reads as a graveyard of one active file (BOOT.md) plus a long tail of +session-scoped notes from March–May 2026 that were never archived or pruned. + +--- + +## STATUS_BOARD counts + In-PR rows + +`STATUS_BOARD.md` is 1854 lines, organized into **89 `## ` sections**, each +with its own `| D-id | deliverable | status |`-shaped table (though the column count +varies per section — some sections add LOC/Priority/Owner columns, which defeats a +single mechanical "grab the last pipe-field" parse). Mechanically extracted: **592 +data rows** (excluding the ~12 repeated header rows). + +**Best-effort status classification** (leading-token match on the status cell, after +stripping bold markers; ~427/592 rows carry a rich narrative status instead of a clean +lifecycle word, which is itself a finding — see Candidate epiphanies): + +| Status (leading token) | Count | +|---|---| +| Narrative / not a clean lifecycle token | 427 | +| Queued | 77 (+1 mentioned mid-cell) | +| Shipped | 33 (+8 mentioned mid-cell) | +| Blocked | 23 (+6 mentioned mid-cell) | +| In PR | 5 (leading-token match only — see caveat below) | +| Proposed | 5 | +| Decision | 3 | +| Superseded | 1 (+1 mentioned) | +| Withdrawn | 1 (+1 mentioned) | + +**Caveat on "In PR" count:** the leading-token classifier only reliably catches rows +using the strict 3-column `| D-id | deliverable | status |` shape. Sections with wider +tables (LOC/Priority/Owner columns) put "In PR" in a MIDDLE column, so the raw text +search for `In PR` anywhere on a line returns **~90 distinct occurrences** across the +file, spanning D-ids from `D-LNC-*`/`D-MW-P2` (2026-09) all the way back to `D-MBX-*`/ +`D-WIKI-HHTL-*`/`D-IDENTITY-2` (2026-06). A fully precise per-row status census was not +attempted at this scale — see Uncertain. + +**Rows whose status text is `In PR` (with the PR number cited) and merge-status check:** + +| D-id | PR cited | `git log --grep` result | +|---|---|---| +| D-LNC-5a, D-MW-P2 (STATUS_BOARD.md:19-20) | #1198 | **MERGED** — `3797237b Merge pull request #1198` exists. **These two rows are STALE**: the board still reads `**In PR** #1198` although #1198 is merged (see Candidate epiphanies). | +| D-GUARD-1 (:802) | #785 | MERGED — `650f925a Merge pull request #785` | +| D-CSW-0 (:803) | #777 | MERGED — `5ce46705 Merge pull request #777` | +| D-CCF-1 (:947) | #628 | MERGED — `6858118b Merge pull request #628` | +| D-CCF-2 (:948) | OGAR #147 | external repo — not checkable from this checkout | +| D-CCF-3 (:949) | q2 #71 | external repo — not checkable from this checkout | +| D-CSV-5b/6b/15 (:1557,1559,1586) | #390 (Wave G tags) | no `Merge pull request #390` commit found in this repo's history — could be a squash-merge with different wording, or an internal wave/work-item tag rather than a GitHub PR number; **inconclusive** | +| D-CSV-10 (:1568) | #388 | same as above — inconclusive | +| D-MBX-A6-P3a/M1 (:1661,1666) | #439 | no direct "Merge pull request #439" found; a later commit (`8017ca19`) DOES reference "D-PG-6" work after #438, suggesting #439 likely landed, but not confirmed by an exact merge-commit grep | +| D-WIKI-HHTL-1/2 (:1672-1673) | #441 | no direct "Merge pull request #441" found; a later fix commit explicitly says `fix(#441): CodeRabbit — STATUS_BOARD #440->#441`, strongly suggesting #441 did land, but not confirmed by an exact merge-commit grep | + +--- + +## ISSUES open + +`ISSUES.md` is 2777 lines, **47 double-entry headers** (`## ISS- () — `, +7 of which put a descriptive title after the id instead of a clean status token in the +header itself — resolved by reading each body for its actual verdict). + +**Counts:** OPEN **37** · RESOLVED **8** · RESOLVING **1** · SUPERSEDED **1** (total 47). + +**All 37 OPEN issues** (id + one-line title, from the header text or, where the header +carries no title, a title derived from the body): + +1. `ISS-NO-NON-LINUX-TARGET` (2026-09-06) — `posix_fadvise` gated `#[cfg(unix)]` but is Linux/Android-only, breaks macOS build +2. `ISS-NO-AARCH64-RUNNER` (2026-09-06) — no CI job ever compiles an aarch64 target; two real defects found only by an external fleet +3. `ISS-EXCLUDED-CRATES-UNBUILT` (2026-09-06) — workspace-excluded crates get zero CI coverage +4. `ISS-PIN-RULING-PROSE-DRIFTS-BEHIND-THE-MANIFEST` (2026-09-05) — pin-rule prose in CLAUDE.md lags the actual Cargo.toml state +5. `ISS-PLAN-TRACKING-IS-UNENFORCED` (2026-09-03) — plans can mint D-ids the board never tracks +6. `ISS-F32-ENGINE-NEVER-CONVERGES-BY-ITS-OWN-DELTA-THRESHOLD` (2026-09-03) — `F32ThinkingEngine::cycle()` never trips its own convergence early-exit; runs full budget every time +7. `ISS-SUPERSESSION-GENERATOR-DID-RANGE-NARROW` (2026-08-28) — `supersession_index.py` only extracts the first id of a `D-XXX-0..4` range +8. `ISS-TOKEN-TENANT-16-COLLIDES-WITH-HOLEV3` (2026-08-28) — two tenants both claim `ValueTenant` discriminant 16 +9. `ISS-TVT-3-DISABLE-RUN-IS-VACUOUS` (2026-08-28) — `token-value-tenant-v1`'s F-TVT-3 disable-run cannot actually fail +10. `ISS-TVT-HYDRATION-REGION-CLAIM-WRONG` (2026-08-28) — plan text contradicts the code it cites (`AWS_DEFAULT_REGION` is optional, plan says required) +11. `ISS-KANBAN-PLAN-EXIT-HAS-NO-NAMED-ROUTE` (2026-08-28) — `KanbanColumn::Plan` is DAG-legal but no primitive ever emits it +12. `ISS-F-MUL-6-HALF-BUILT` (2026-08-27) — only one of two known consumers was actually build-verified +13. `ISS-MUL-GATE-NAMED-FOR-THE-WRONG-LAYER` (2026-08-27) — `contract::mul::GateDecision` is really an execution/commit gate, module name misleads +14. `ISS-PLANNER-SANDBOX-STILL-CARRIES-FREE-TEXT` (2026-08-27) — `MulGateDecision::Sandbox{reason:String}` still heap-allocates free text +15. `ISS-NO-CAUSAL-SIGN-ON-EDGES` (2026-08-26) — `CausalEdge64` has no Inc/Dec polarity carrier +16. `ISS-D-ECG-6-BUDGET-WITHOUT-ADMISSION` (2026-08-26) — census-rank frontier has no relevance-ADMIT stage in front of it +17. `ISS-RUNG-VS-BAND-CARDINALITY-COLLISION` (2026-08-26) — `RungLevel` (10 values) vs `ReasoningBand` (8 values) share endpoint names, differ in cardinality +18. `ISS-ALPHA-NOT-LOAD-BEARING` (2026-08-26) — `AttentionMaskSoA` has no production consumer at all +19. `ISS-REASONING-BAND-GATES-NOTHING` (2026-08-26) — `ReasoningBand` is minted/tested but gates no control-loop behaviour +20. `ISS-BAND-READING-UNMINTED-IN-OGAR` (2026-08-26) — the OGAR producer side of `ClassView::band_reading` never answers +21. `ISS-DOMAIN-LENS-BY-CONVENTION-ONLY` (2026-08-26) — a real domain-conditioned-lens ruling exists only as prose in one consumer +22. `ISS-MUL-GATE-OUTCOME-COUPLED-TO-PRODUCER-GROUND` (2026-08-26) — `GateDecision::{Hold,Block}` narrowed to MUL-specific vocabulary at a public boundary +23. `ISS-PERTURBATION-P64-ADDRESS-IDENTITY-UNPROVEN` (2026-08-26) +24. `ISS-CORPUS-ADDRESSING-OPEN-POINTS` (2026-08-22) +25. `ISS-CAUSAL-EDGE-CARRIES-SEVEN-PRE-EXISTING-CLIPPY-FINDINGS` (2026-08-22) +26. `ISS-HYDRATE-ENV-READER-IS-A-SECOND-COPY-OF-DEV-S3-ENV` (2026-08-17) — deliberate, has a named exit +27. `ISS-HYDRATE-NAME-COLLIDES-WITH-TWO-EXISTING-WORKSPACE-MEANINGS` (2026-08-17) — cosmetic, no correctness impact +28. `ISS-HELIX-GOLDEN-STEP-LABEL` (2026-08-12) — misleading label, not wrong code +29. `ISS-D-IGN-B-REAL-CORPUS-PATH-IS-UNVERIFIED` (2026-08-06) +30. `ISS-IDENTITY-CODEBOOK-ORDINAL-STABILITY` (2026-08-06) — partially mitigated +31. `ISS-REMOTE-URI-CONSTRUCTORS-PREDATE-THE-HYDRATION-DOCTRINE` (2026-08-06) +32. `ISS-CODEC-RESEARCH-MDCT-ASSERT` (2026-08-05) — pre-existing, discovered not caused +33. `ISS-MARM-T1-4X-A0-GAP` (2026-08-05) — measurement defect, not a result +34. `ISS-MAILBOXSOA-ROW-COST-VS-512B-CANON` (2026-08-04) — an open question, not yet a conclusion +35. `ISS-DOMINO-WRITES-ENERGY-OUTSIDE-ITS-OWN-SCHEMA` (2026-07-29) +36. `ISS-NO-PER-THREAD-TEMPORAL-PROJECTION-IS-EVER-CONSTRUCTED` (2026-07-29) — upstream of `meta_basin` +37. `ISS-CLASSID-OGAR-DRIFT` (2026-06-20 entry, marked OPEN — needs operator sign-off; note a LATER dated entry at line 2771 for the same id is marked RESOLVING/landed, i.e. this id has two append-only dated sub-entries with different statuses — the most recent one (RESOLVING) likely supersedes this OPEN one in practice) + +--- + +## TECH_DEBT counts + open P0/P1 + +`TECH_DEBT.md` is 4311 lines and uses **at least three different header conventions** +across its history (`## TD-... — OPEN`, `### TD-...` with a `**Severity:**`/`**Open.**` +body, and `## YYYY-MM-DD — TD-...:` date-prefixed) — itself a finding (see Candidate +epiphanies). Unified parse across all three: **167 total entries**. + +**Status counts** (header-token first, else body `**Status:**`/`**Open.**` marker, else +UNLABELED): + +| Status | Count | +|---|---| +| OPEN | 95 | +| UNLABELED (no explicit status found in header or body) | 65 | +| PAID | 4 | +| RESOLVED | 3 | + +The 65 UNLABELED entries are concentrated in the oldest part of the file (PR #270-#330 +era items like `TD-F10-ACTOR-ID`, `TD-ARROW-58`, `TD-INT-1`..`TD-INT-14`) which predate +the "Kanban Format (priority + scope on every entry)" convention introduced at line 1964 +— they were never retrofitted with the newer schema. + +**Entries tagged Priority/Severity P0 or P1 that are still Open (19 of 20 P0/P1-tagged +entries; the 20th, `TD-INT-1/2/4`, is tagged Paid):** + +| Line | Priority | TD-id | +|---|---|---| +| 1758 | P1 | `TD-NDARRAY-SIMD-UNPACK-I4-16D` (W1a-#1) | +| 1772 | P1 | `TD-NDARRAY-SIMD-SATURATING-ABS-I8` (W1a-#2) | +| 1786 | P1 | `TD-NDARRAY-SIMD-GATHER` (W1a-#3) | +| 1813 | P1 | `TD-NDARRAY-SIMD-POPCOUNT-U64` (W1a-#5) | +| 1897 | **P0** | `TD-SIMD-SWEEP-W4` (lance-graph-contract mul.rs follow-up) | +| 1980 | P1 | `TD-ARIGRAPH-EPISODIC-FIDELITY-1` — AriGraph episodic retrieval transcoded as the RAG baseline the paper beats | +| 2055 | P1 | `TD-Q2-STUBS-DEDUP-1` — q2 carries local lance-graph/ndarray stubs needing re-export | +| 2116 | **P0** | `TD-API-DRIFT-MIDFLIGHT-1` — consumer migrations failing mid-air on source-crate API drift | +| 2159 | P1 | `TD-SUPER-DOMAIN-SUBCRATES-1` | +| 2242 | P1 | `TD-THINKING-ENGINE-UNWIRED-1` — 582 KB cognitive substrate dormant | +| 2289 | **P0** | `TD-SDR-PR-FOLLOWUP-1` — 5 commits stacked on merged main, no follow-up PR opened | +| 2309 | **P0** | `TD-SDR-CONSUMER-PUSH-1` — medcare-rs/smb-office-rs bridge wirings committed LOCALLY, not pushed | +| 2327 | P1 | `TD-SDR-AUDIT-PERSIST-1` — audit events emit to in-memory chain only, no persistent sink | +| 2347 | P1 | `TD-SDR-FAMILY-HYDRATION-1` | +| 4018 | P1 | `TD-CONTEXT-BUNDLE-2` | +| 4048 | P1 | `TD-GENERIC-BRIDGE-3` | +| 4069 | P1 | `TD-MANIFEST-MODULES-4` | +| 4097 | P1 | `TD-RACTOR-SUPERVISOR-5` | +| 3995 | **P0** | `TD-OGIT-G-SLOT-1` — wire u32 G slot into the SPO quad-store | + +5 of these are P0 (`TD-SIMD-SWEEP-W4`, `TD-API-DRIFT-MIDFLIGHT-1`, `TD-SDR-PR-FOLLOWUP-1`, +`TD-SDR-CONSUMER-PUSH-1`, `TD-OGIT-G-SLOT-1`); 14 are P1. + +--- + +## INTEGRATION_PLANS index vs directory + +`.claude/plans/` holds **211 files**. `INTEGRATION_PLANS.md` cites **141 distinct +`.claude/plans/*.md` paths** by literal path string. + +- **Plans that exist in `.claude/plans/` but are NOT cited by path in + `INTEGRATION_PLANS.md`: 72.** Of those, 14 are mentioned by bare filename (without the + full `.claude/plans/` path) somewhere in the index text — likely still "known" to the + index in spirit. **58 plans are mentioned nowhere in `INTEGRATION_PLANS.md` at all** + (neither full path nor bare name) — see the list below. A large sub-cluster (20 files) + is the entire `3DGS-*` doc family (has its own `3DGS-PLAN-INDEX.md` sub-index instead), + and another sub-cluster is the `tesseract-rs-*` transcode plans (5 files) and the + `alpha-reason-witness-shader-field-*` archaeology plans (3 files). + + Full 58-item unmentioned list: `3DGS-3D-Tiles-runtime-plan`, `3DGS-4x4-cognitive-shader-integration-plan`, + `3DGS-ArcGIS-Cesium-ingestion-plan`* , `3DGS-Blender-transcode-crosspollination-plan`, + `3DGS-Cesium-BindSpace4-headstone-exploration`, `3DGS-Cesium-feature-mapping-plan`, + `3DGS-HHTL-datalake-traversal-plan`, `3DGS-Lance-Arrow-storage-plan`, `3DGS-PLAN-INDEX`, + `3DGS-PRX12-crosspollination-capstone`, `3DGS-SplatShaderBlas-BLASGraph-crosspollination-plan`, + `3DGS-blast-radius-application-map`, `3DGS-certified-query-render-plan`, + `3DGS-cross-pollination-raw-field-plan`, `3DGS-domain-adapter-strategy-plan`, + `3DGS-epiphany-roadmap-plan`, `3DGS-genetics-4x4-fanout-plan`, `3DGS-integration-wiring-plan`, + `3DGS-neuronal-network-4x4-plan`, `3DGS-ultrasound-SaMD-plan`, + `Palette256-3DSB-PhiSpiral-attention-integration-plan`, `PhiSpiral256-SoA-cross-system-integration-plan`, + `a3-carrier-v1`, `alpha-reason-witness-shader-field-archaeology-pass-1`, + `alpha-reason-witness-shader-field-d-arw-0-audit-v1`, `alpha-reason-witness-shader-field-lineage-addendum-v1`, + `archetype-scaffold-v1`, `belief-abi-restoration-v1`, `burn-ndarray-parity-sprint-v1`, + `cognitive-substrate-convergence-v2`, `cognitive-substrate-convergence-v3`, + `cognitive-write-roundtrip-substrate-v1`, `deepnsm-morton-comma-facet-v1`, `dialectic-engine-v1`, + `dtsc1-thinkingstyle-dedup-spec-v1`, `epistemic-quadrant-materialization-v1`, + `genetic-research-substrate-integration-v1`, `genetics-probes-v1`, `hydrate-crate-hardening-council-v1`, + `lance-graph-rdf-fma-snomed-v1`, `lance9-datafusion54-upgrade-probe-v1`, `measure-64k-axes-v4`, + `ocr-canonical-soa-integration-v1`, `ocr-probes-v1`, `oracle-funnel-probe-v1`, + `oxigraph-arigraph-cognitive-shader-soa-merge-v1`, `post-438-integration-options-v1`, + `probe-revision-attention-view-1`, `reliability-checklist-arc-v1`, `soa-migration-diff-resolution-2026-06-13`, + `sprint-5-through-9-roadmap-v1`, `sql-spo-ontology-bridge-v1`, `tesseract-rs-ast-dll-codegen-v1`, + `tesseract-rs-layout-transcode-v1`, `tesseract-rs-recodebeam-transcode-v1`, + `tesseract-rs-traineddata-gguf-v1`, `tesseract-rs-transcode-master-v1`, + `tetrahedral-epiphany-splat-integration-v1`, `transcode-extend-core-probe-v1`, + `wikidata-lazy-spine-hydration-v1` + +- **Paths cited in `INTEGRATION_PLANS.md` that do NOT exist as a file in `.claude/plans/`: 2.** + 1. `.claude/plans/pdf-to-text-ocr-v1.md` — also referenced from `tesseract-rs`'s own + `CLAUDE.md` ("plan `pdf-to-text-ocr-v1.md`"); may genuinely live only in the + `tesseract-rs` repo, or the reference is stale in both places. + 2. `.claude/plans/splat-native-ultrasound-simd-substrate-v1.md` — the directory has + `splat-native-ultrasound-v1.md` (no `-simd-substrate` infix); looks like a naming + drift/typo between the citation and the actual filename. + +--- + +## LATEST_STATE summary + +`LATEST_STATE.md` is 3460 lines, read across multiple offsets (structure: newest-first +append-only dated delta entries at the top, running from **2026-09-07** back through +2026-05-05, followed by several STATIC/summary sections near the bottom of the file that +have not been kept in sync with the dated deltas above them). + +- **"as of" date:** **2026-09-07** (top entry: `D-BLOCKS-HOTPLUG-1`, an operator ruling + that a hot-plugged consumer classid is not a canon builtin). + +- **"Current Contract Inventory (lance-graph-contract)" section** (line 2915-2982): + **27 blockquote (`> **...`) entries**, spanning dates 2026-05-30 through 2026-06-26. + This is a STATIC section — it has NOT been updated with any of the ~100+ dated delta + entries that were prepended above it since 2026-06-26 (over two months of newer + contract changes are not reflected here). See Candidate epiphanies. + +- **"Immediate Next Work" / Queued Work section** (verbatim, lines 3073-3101): + + > **Queued Work — sprint-13 (specs being drafted in the sprint-13-preflight fleet on + > this branch):** + > - **D-CSV-13b** — SIMD vectorization of D-CSV-8 i4 MUL evaluation. **IN PR + > (sprint-13/W-I1 salvage)**... + > - **D-CSV-14** — on-Think method migration for D-CSV-12 splat ops... + > - **D-CSV-16** — NEW sprint-13 entry. Spec being drafted by PP-5. + > - **D-CSV-17** — NEW sprint-13 entry. Spec being drafted by PP-3. + > + > **Sprint-14+ Phase F items (Backlog):** + > - ndarray `parallel`-feature `par_*` rayon variants for `QualiaStream` / + > `InferenceStream` / `SplatFieldStream` (work-stealing). + > - D-REUNIFY-4/5/6 carryover from causaledge64-mailbox-rename-soa-v1... + > + > **`codec-sweep-via-lab-infra-v1` Phase 0 remainder (carry-over):** + > - D0.1, D0.2, D0.3, D0.5 + four test gates (listed verbatim in the file). + > + > **`elegant-herding-rocket-v1` Phase 2 (still queued):** + > - D2, D3, D5, D7 (listed verbatim in the file). + + **This entire section is dated to sprint-13/sprint-14 (~2026-05-16) and is STALE**: + D-CSV-13b is cited here as "IN PR" but STATUS_BOARD.md's own D-CSV-13b row (line 1584) + shows it long since shipped with measured SIMD benchmarks. This "Immediate Next Work" + section has not been refreshed to reflect the file's own top-of-file 2026-09-07 state. + +- **"Deferred (do NOT propose these — they're explicitly parked)" section** (verbatim, + lines 3102-3119): + + > **Sprint-12/13 explicit deferrals (2026-05-16):** + > - **TD-COLLAPSE-GATE-SMALLVEC-1** — CLOSED 2026-06-11 as moot: `CollapseGateEmission` + > removed entirely (PR #477 tombstone commit), nothing left to optimize. + > - **TD-SIGMA-TIER-THRESHOLDS-1** — Σ10 VAMPE-coupled Jirak-derived threshold + > refinement (D-CSV-15). Hand-tuned acceptable through sprint-12 per + > `I-NOISE-FLOOR-JIRAK`; principled Jirak 2016 derivation forwarded to sprint-13+ + > VAMPE coupled-revival track. + > - **ndarray `parallel`-feature `par_*` rayon variants** — productized substrate + > ships sequentially in PR #147; rayon work-stealing wraps deferred to sprint-14+ + > behind an opt-in feature gate. + > + > **Long-running parks (pre-existing):** + > - CausalityFlow TEKAMOLO extension (modal/local/instrument + beneficiary/goal/source, + > 9 total) — struct change deferred until after Phase 2. + > - D8 story-context bridge, D9 ONNX arc export, D10 Animal Farm validation, D11 + > bundle-perturb emergence — Phase 3/4. + > - Named Entity pre-pass (NER) — biggest OSINT blocker, separate PR. + > - FP_WORDS = 160 migration (currently 157) — coordinated ndarray change. + > - Crystal4K 41:1 persistence compression. + > - 200-500 YAML TEKAMOLO templates per language — future training pipeline. + > - Python/TypeScript grammar-stack convergence. + + One entry in this list (`TD-COLLAPSE-GATE-SMALLVEC-1`) is self-annotated CLOSED as of + 2026-06-11, i.e. this "Deferred" list already contains a stale row it did not remove. + +--- + +## IDEAS current list + +`IDEAS.md` is 1236 lines, **47 `## ` headers total** — a mix of structural/instructional +headers (Triple-entry discipline, Governance, Kanban Format, Open Ideas, Implemented +Ideas, Rejected/Deferred Ideas, How-to-use, template placeholders) and **33 actual dated +idea entries**. Status extraction (via `**Status:**` line where present): + +| Line | Date | Status | Title | +|---|---|---|---| +| 90 | 2026-09-05 | Open | Σ-propagation: the batched `F64x8` kernel is the real lever; AMX/MKL was the wrong shape for a 2×2 | +| 98 | 2026-09-05 | Open | The σ column indexes a codebook that exists in three doc comments and zero crates | +| 106 | 2026-09-05 | Open | `ndarray::simd::ternlog` ⋂ the §14 voxel cube: the "other 18 cells" are one named immediate each | +| 114 | 2026-09-05 | Open | Citations should carry the path the decay gate can resolve, or the gate sees 7% of them | +| 122 | 2026-06-15 | RESEARCH | CLAM residue ladder ⋂ knee/hip attractor basins (HHTL cascade in REVERSE) | +| 140 | 2026-05-13 | (none found) | CORRECTION-OF previous same-day splat row: split into two distinct ideas | +| 144 | 2026-05-13 | (none found) | EXECUTION PATH: prerendered cinematic as palette-indexed frames in LanceDB | +| 206 | 2026-05-13 | (none found) | REFRAME: holographic cinematic is a SALES asset, not a product feature | +| 230 | 2026-05-13 | (none found) | RECONCILIATION: Amiga-demoscene prerender + live 60fps renderer compose, not compete | +| 250 | 2026-05-13 | (none found) | Sci-fi presentation vision: transparent holographic human-body projection for q2 | +| 278 | 2026-05-13 | (none found) | CORRECTION-OF prerender row: ndarray already ships the 60fps double-buffer renderer | +| 282 | 2026-05-13 | (none found) | Separate-and-orthogonal: 3D Gaussian-Splat prerender buffer as Tier-3 FMA render path | +| 286 | 2026-05-13 | Open | Super-domain subcrate scaffolding cascade (MedCare→smb-bridge→woa-rs→hiro/hubspot) | +| 332 | 2026-05-13 | Open | Pattern E+F+cognition cascade: manifest + ractor supervisor + thinking-engine bridge | +| 365 | 2026-05-13 | Open | Wire `thinking-engine` into UnifiedBridge — collapse D-SDR-13/15/17 into one module | +| 414 | 2026-04-29 | Implemented | Probe P1: γ-phase-offset ranking discrimination (from 2026-04-29) | +| 499 | 2026-04-29 | Open | Inverted-pyramid awareness streaming via CausalEdge64 through SPO+COCA→CAM_PQ | +| 522 | 2026-04-29 | Implemented | Probe P1: γ-phase-offset ranking discrimination | +| 548 | 2026-04-29 | Open | Safetensor-Streaming als ndimensionale Bedeutungsakkumulation | +| 566 | 2026-04-29 | Open | Family-Bounds als globale fraktale Codierung (Hypothesis Test) | +| 583 | 2026-04-29 | Open | Pillar 7 Front-to-Back α-Akkumulation (LIKELY-REDISCOVERY) | +| 600 | 2026-04-29 | Open | Pillar 8 Adaptive Densification für Σ-Codebook | +| 615 | 2026-04-29 | Open | Pillar 9 SH-Koeffizienten als Thinking-Style-Manifold | +| 632 | 2026-04-19 | Open | FP_WORDS = 256 (supersede the 160 plan) | +| 663 | 2026-04-19 | Open | CORRECTION-OF FP_WORDS = 256 | +| 721 | 2026-04-19 | Open | REFINEMENT-OF CORRECTION-OF FP_WORDS scope | +| 772 | 2026-04-19 | Open | REFINEMENT-2: HDC substrate is FP16/BF16, not FP32 | +| 829 | 2026-04-19 | Open | lance-graph-cognitive refactor: dedup + merge + excise | +| 886 | 2026-04-19 | Open | CORRECTION-OF lance-graph-cognitive refactor | +| 904 | 2026-04-19 | Open | Fractal round-trip codec: phase+magnitude preservation | +| 957 | 2026-04-19 | Open | Fractal codec validation path: codec_rnd_bench + ICC_3_1 | +| 1006 | 2026-04-19 | Open | Zipper codec: phase+magnitude multiplexed in one bgz17 container | +| 1088 | 2026-05-05 | Reshaped | Future-work items extracted from PRs #244–#335 | + +Structural note: entries at lines 499-1006 sit AFTER the file's own "## How to use this +file" section (line 479) — an unusual placement for an append-only-at-top idea log; not +fully explained within budget (see Uncertain). + +--- + +## EPIPHANIES recent + id checks + +`EPIPHANIES.md` is **27,617 lines** — not read whole per task instructions; only +`grep -n "^## "` used. + +**15 most recent headings (verbatim, newest-first):** + +1. `## 2026-09-07 — E-PLUG-AND-PLAY-IS-THE-DECLARATION-NOT-A-TABLE-1 — my fix rebuilt the lockstep it was closing` +2. `## 2026-09-07 — E-A-DOC-COMMENT-IS-NOT-A-FAIL-CLOSED-MECHANISM-1 — hotplug could still land on V1 with a one-liner` +3. `## 2026-09-07 — E-A-V3-MINT-MUST-NEVER-DEGRADE-TO-V1-1 — the fallback arm's own justification was falsified by D-BLOCKS-HOTPLUG-1` +4. `## 2026-09-07 — E-THE-V1-GUARD-WAS-TESTED-THE-V3-GUARD-THAT-REPLACED-IT-WAS-NOT-1 — a guard nothing proves can fire is the defect one level up` +5. `## 2026-09-06 — E-AN-EXCLUDED-CRATE-ON-AN-X86-ONLY-FLEET-IS-CODE-NO-CI-HAS-EVER-COMPILED-1 — un-gating one downstream suite found a second aarch64 defect that could never have built` +6. `## 2026-09-06 — E-I-CITED-THE-RIGHTMOST-REGISTER-AND-CALLED-IT-THE-ADDRESS-1 — three corrections to one entry, each because I reasoned instead of measuring` +7. `## 2026-09-05 — E-A-SWEEP-IS-COMPLETE-ONLY-WITHIN-THE-TARGET-KINDS-ITS-GATE-COMPILES-1 — #1194 swept the whole 1.98 delta and still left two sites, because "whole" was measured through six clippy steps` +8. `## 2026-09-06 — E-THE-AARCH64-PATH-HAD-NEVER-BEEN-COMPILED-1 — a cfg-gated arch path is dead code until some CI targets it` +9. `## 2026-09-05 — E-A-MACHINE-APPLICABLE-FIX-IS-A-SUGGESTION-NOT-A-PROOF-1 — clippy's own autofix did not compile, and the lint it fixes is a substrate argument` +10. `## 2026-09-05 — E-PLANNING-MIGRATES-TO-LOCO-R2IL-DATAFUSION-IS-GRACE-PERIOD-1 (OPERATOR-RULED)` +11. `## 2026-09-05 — E-THE-UNFINISHED-UDF-WAS-NOT-THE-DEBT-1 — D-OIF-1 re-derived: the execution model that needed policy_hash_v1 never reached production` +12. `## 2026-09-05 — E-BLW5-FIRST-MEASUREMENT-1 — the observer-effect instrument is alive, and every pre-registered gate reads SILENT` +13. `## 2026-09-05 — E-NARS-EXPECTATION-CHOICE-PREFERS-IGNORANCE-TO-A-CONFIDENT-NEGATIVE-1 — a derived belief with f < 0.5 loses CHOICE to a vacuous one` +14. `## 2026-09-05 — E-THE-UNFINISHED-FUNCTION-WAS-NOT-THE-DEBT-1 — the execution model that needed policy_hash_v1 never reached a binary, and neither has its replacement` +15. `## 2026-09-05 — E-A-COLUMN-OF-INDICES-INTO-A-CODEBOOK-THAT-DOES-NOT-EXIST-1 — three stale idea cards, each wrong about its own blocker` + +(Note: item 8 dated 2026-09-06 sits chronologically AFTER item 7 dated 2026-09-05 in the +file's own prepend order — the append-only ledger is not strictly date-sorted at every +boundary, likely from two branches merging in close succession.) + +**Requested id existence checks:** + +| Id | Exists? | Location | +|---|---|---| +| `E-SIGNATURE-PDE-SWEEP-SHIPPED-W1.5-GATE-WAS-QUIETLY-OPEN-1` | **YES** | line 2195, dated 2026-09-02 | +| `E-PILLAR-11-PUBLISHED-BOUND-NEEDS-ITS-OWN-NUMERIC-GUARD-1` | **YES** | line 2253, dated 2026-09-02 | +| `E-PLANNING-MIGRATES-TO-LOCO-R2IL-DATAFUSION-IS-GRACE-PERIOD-1` | **YES** | line 539, dated 2026-09-05 (operator-ruled) | +| `E-LANCE-IS-UPSTREAM-AUTHORITATIVE-1` | **YES** | line 12295, dated 2026-08-05 (operator-ruled); also referenced elsewhere at line 583 | + +**`log_signature` / `LOG-SIGNATURE` / `LYNDON` headings:** one substantial entry found — +`## 2026-09-03 — E-A-DOCUMENTED-MODULE-THAT-WAS-NEVER-\`pub mod\`-D-IS-DEAD-CODE-WITH-A-PARITY-CLAIM-1` +(line 2094) — `crates/sigker/src/log_signature.rs` existed on `main`, 388 lines, cited a +Reizenstein-Graham paper and a "7-13× compression, lossless" claim, but `lib.rs` never +declared `pub mod log_signature;` — it had never been compiled once. The entry also +corrects a basis-projection bug (flat read ≠ true Lyndon-basis coefficients) and +corrects the compression ratio (2.10×-7.62× depending on depth, not a flat "7-13×"). +`TD-NDARRAY-SIMD-LYNDON-PACK` (TECH_DEBT.md line 1854, tagged P3/DEFERRED) is the +still-open SIMD follow-up this entry unblocks but does not itself resolve. + +--- + +## SUPERSESSION-INDEX summary + +`SUPERSESSION-INDEX.md` is 159 lines, fully read (it is generated + small). + +- **Row counts:** Table 1 (ruled symbols) = **14 rows**. Table 2 (plans naming a ruled + symbol without citing the ruling) = **73 rows**, broken down as **RESCOPE: 55 · READ: + 18 · ARCHIVE?: 0**. + +- **14 ruled symbols, with verdict:** + + | Symbol | Verdict | Successor | Live in crates | Named in plans | Blind plans | + |---|---|---|---|---|---| + | `A2AMessage` | BLOCKED | — | 2 | 1 | 0 | + | `StepMask` | BLOCKED | — | 3 | 9 | 4 | + | `commit_to_l4` | BLOCKED | — | 2 | 2 | 0 | + | `dispatch_busdto` | BLOCKED | — | 3 | 8 | 5 | + | `persist_cycle` | BLOCKED | — | 11 | 8 | 5 | + | `CognitiveMarkers` | REPURPOSE | `Commit` | 1 | 0 | 0 | + | `DominoCascade` | REPURPOSE | `Commit` | 7 | 1 | 0 | + | `GateDecision` | REPURPOSE | — | 25 | 27 | 24 | + | `GateState` | REPURPOSE | — | 14 | 1 | 1 | + | `MergeMode` | REPURPOSE | — | 8 | 13 | 12 | + | `ResonanceDto` | REPURPOSE | `PerturbationDto` | 2 | 12 | 7 | + | `BindSpace` | **RETIRE** | — | 68 | 47 | 41 | + | `CollapseGateEmission` | **RETIRE** | — | 5 | 14 | 12 | + | `ThinkingStyle` | RETIRE-toward-contract | — | 52 | 28 | 24 | + + Headline (the file's own commentary, verified against the table): `BindSpace` is + marked RETIRE and is simultaneously the most-referenced symbol in the whole index — + 68 crate files, 47 plans, 41 of them "blind" (naming it without citing the ruling). + +- **73 plans flagged as naming a ruled symbol without citing the ruling** — the full + 55-item RESCOPE list and 18-item READ list are reproduced verbatim in the file at + `.claude/board/SUPERSESSION-INDEX.md:82-154`; not re-transcribed here for space, but + every row is present in the file (read in full, see above). Two rows worth flagging + individually: `GateDecision` review found the mechanical `ARCHIVE?` route's first + batch was **3/3 false positives** (the file documents this itself, lines 73-78) — so + the route is explicitly self-described as "a prompt to read the plan, never a licence + to act on it," and this session did not attempt to re-verify any individual RESCOPE/ + READ row's current correctness (that would require reading up to 55+18 plan files). + +--- + +## Candidate epiphanies + +Facts observed this pass that look like genuine board-hygiene findings, each with a +file:line anchor, offered as raw material — not claimed as ratified epiphanies. + +1. **STATUS_BOARD.md:19-20** — `D-LNC-5a` and `D-MW-P2` both read `**In PR** #1198`, but + `#1198` has a merge commit on `main` (`3797237b`, verified via + `git log --oneline --all --grep="Merge pull request #1198"`). These two rows are + stale post-merge — the board hygiene rule (root `CLAUDE.md` "Mandatory Board-Hygiene + Rule") requires a merged-PR row to update in the same commit; here it evidently did + not, or a later merge (of #1199, which the AGENT_LOG shows recording #1198's own + merge — `.claude/board/LATEST_STATE.md:213-227`) never touched these two rows. + +2. **LATEST_STATE.md:2915-2982 vs :1-2914** — the "Current Contract Inventory" section + (27 entries, frozen at 2026-06-26) has not been updated in over two months of newer + dated deltas prepended above it (through 2026-09-07). A session reading only the + named section header would get a badly stale contract inventory; the real current + state is scattered across ~100+ dated top-of-file entries instead. + +3. **LATEST_STATE.md:3073-3119** — "Immediate Next Work" / "Deferred" sections are dated + internally to sprint-13/sprint-14 (~2026-05-16) and cite work (e.g. D-CSV-13b "IN PR") + that STATUS_BOARD.md's own row for the same D-id (line 1584) shows shipped with + measured benchmarks long since. One row in "Deferred" (`TD-COLLAPSE-GATE-SMALLVEC-1`) + is even self-annotated "CLOSED 2026-06-11 as moot" while still sitting in the + "Deferred (do NOT propose these)" list. + +4. **TECH_DEBT.md** — the file uses at least 3 distinct header conventions across its + history (`## TD-X — OPEN`, `### TD-X` + body `**Severity:**`/`**Open.**`, and + `## YYYY-MM-DD — TD-X:`), and the "Kanban Format (priority + scope on every entry)" + convention (line 1964) was introduced partway through the file's life — 65 of 167 + entries (39%) carry no machine-extractable status at all, mostly the oldest + PR-numbered items (`TD-F10-ACTOR-ID` .. `TD-INT-14`, lines 3513-4264) that predate the + convention and were never retrofitted. + +5. **CLAUDE.md** (root) "Prior art" section says "`.claude/*.md` (61 top-level docs)" — + the actual count today is 71. The section also names `SESSION_CAPSTONE.md` as an + example file, which does not exist anywhere in this repo (`ls .claude/*.md` has no + such file) — a dangling citation in the repo's own operational-contract doc. + +6. **`.claude/pattern.md` vs `.claude/patterns.md`** — near-duplicate purpose (both + "SoA/DTO graph traversal" primers, both claim "read by every session" / "READ FIRST" + before touching that surface), both still actively cross-referenced (17 and 30 + internal refs respectively) and both younger than most of the directory (2026-05-06 + and 2026-05-12) — worth a session asking whether these should be merged rather than + maintained as two parallel entry points to the same doctrine. + +7. **INTEGRATION_PLANS.md coverage gap** — 58 plan files (27% of the 211 in + `.claude/plans/`) are named nowhere in the index, concentrated in two coherent + clusters (`3DGS-*`, 20 files, tracked instead by its own `3DGS-PLAN-INDEX.md`; and + `tesseract-rs-*` transcode plans, 5 files) — suggests `INTEGRATION_PLANS.md`'s + "index of every plan" framing (root CLAUDE.md: "Consult before proposing a new plan") + is not actually comprehensive for at least these two whole subject areas. + +--- + +## Uncertain + +- **STATUS_BOARD.md "In PR" full census**: given the file mixes 3-column, 4-column and + 6-column table shapes across 89 different plan sections (592 data rows total), a fully + precise single-pass mechanical extraction of "every row whose status is exactly In PR" + was not completed — only the rows matching the strict `| D-id | deliverable | status |` + 3-column shape were reliably classified (5 rows), plus a broader raw-text sweep found + ~90 lines mentioning "In PR" somewhere. The 10 clearest single-D-id "In PR" citations + were individually checked against `git log`; PR numbers #390, #388, #439, #441 could + not be confirmed merged or open via an exact `git log --grep="Merge pull request #N"` + match (no such commit found), which could mean squash-merge wording differs, the + numbers are internal wave tags rather than GitHub PR numbers, or they are genuinely + still open — not resolved within this pass's budget. + +- **TECH_DEBT.md "still open" for the 65 UNLABELED entries**: this report treats + UNLABELED as "status unknown," not as "open." Per the file's own stated policy + ("Debt moves Open → Paid by status-flip; rows are NEVER deleted"), an entry with no + flip marker plausibly IS still open by default, but that inference was not verified + entry-by-entry (would require reading ~65 bodies individually). + +- **IDEAS.md structural placement**: dated entries at lines 499-1006 (2026-04-19 through + 2026-04-29 topics) sit physically AFTER the "## How to use this file" instructional + section (line 479), which is an unusual position for an append-only-at-top idea log. + Not investigated further — could be a historical bulk-import placed below the + instructional boilerplate, or a structural quirk from an earlier file-reorganization; + flagged for a follow-up read rather than resolved here. + +- **ISS-CLASSID-OGAR-DRIFT double entry** (ISSUES.md lines 2771 and 2774): two dated + sub-entries for the same issue id, one OPEN (older position in the file, meaning it + was likely written LATER given the append-at-top convention — but the OPEN one's own + visible date-string is "2026-06-20" same as the RESOLVING one) — the file's own + ordering vs the embedded date strings appear to conflict slightly; treated both as + distinct rows in the OPEN/RESOLVED counts above without fully resolving which is + chronologically authoritative. + +- **7 top-level docs whose header carries a descriptive title instead of a clean + OPEN/RESOLVED-style token** (in ISSUES.md) required reading their bodies to determine + status; this was done for all 7, but a couple (e.g. `ISS-F32-ENGINE-NEVER-CONVERGES...`) + had no single unambiguous "OPEN"/"RESOLVED" sentence — the OPEN classification there + is inferred from prose ("Not root-caused or fixed... Not scheduled") rather than a + literal status field. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-v3-and-harvest.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-v3-and-harvest.md new file mode 100644 index 000000000..2ac736df1 --- /dev/null +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-v3-and-harvest.md @@ -0,0 +1,238 @@ +# V3 + Harvest inventory — lance-graph, 2026-09-07 + +All 21 files under `.claude/v3/` read in full (README, INTEGRATION-PLAN, +COMPONENT-MAP, ENTROPY-MILESTONES, MODULE-TABLE [top + tail sections; the +middle is a 304-row mechanical per-file census, sampled representatively — +not individually re-verified row-by-row], FUTURE-DESIGN, VISION, +agents/BOOT.md, 7 knowledge/*.md, 6 soa_layout/*.md). Note: the task brief +said "22 files" — the actual count on disk is 21; nothing appears missing +(`find .claude/v3 -type f` = 21 files, all read). + +--- + +## v3 file table + +| File | Role (one line) | Last commit | ⊘/SUPERSEDED/RETIRED markers carried | +|---|---|---|---| +| `README.md` | Orientation + doc map, entry point for `/v3` skill | 2026-07-10 | none itself; points at primer §6 for supersession list | +| `INTEGRATION-PLAN.md` | The W0–W6 wave plan + 15 dated addenda (Addendum-1..15) recording live corrections | 2026-08-04 | Addendum-6→7 (mutual-masking retracted), Addendum-14 (ack eliminated), Addendum-15 (⊘ SUPERSEDED note on W1/W2 rows — "rows stay as written, THIS note is current state") | +| `COMPONENT-MAP.md` | Per-subsystem REUSE/REPURPOSE/EXTEND/RETIRE/NEW/BLOCKED verdict table, 7 sections | 2026-09-05 | one inline ⊘ **PARTLY STALE** note on `KanbanActor` (2026-09-05, actor half deleted 2026-08-05) | +| `ENTROPY-MILESTONES.md` | The N→1 duplication-collapse ledger, M1–M26, each with a mechanical gate | 2026-07-23 | M9 row itself carries "RESOLVED... REOPENED (partial)"; several rows are now stale relative to later PRs (see Uncertain) | +| `MODULE-TABLE.md` | Mechanical per-file census (304 files: lance-graph, lance-graph-contract, + an "ancestry pipeline" addendum for thinking-engine/p64-bridge/cognitive-shader-driver) | 2026-07-10 | none as markers; content itself documents 4+ independent 12-entry "style table" duplications not yet swept | +| `FUTURE-DESIGN.md` | APPEND-ONLY meta-board / wiring-queue index for post-2026-07-02 rulings, newest-first | 2026-07-18 | ⊘ READ-THROUGH note on the "ack is the kanban trigger" row (superseded by ack-elimination) | +| `VISION.md` | The graded WHY (every claim tagged [G]/[RULING]/[ASPIRATION]) | 2026-07-17 | none; self-describes as the append-only "vision-keeping" doc | +| `agents/BOOT.md` | V3 agent-card activation table (4 cards) extending main `.claude/agents/BOOT.md` | 2026-07-02 | none | +| `knowledge/v3-substrate-primer.md` | One-page doctrine summary, §6 "what must NOT be reinvented" table | 2026-07-17 | is itself the canonical supersession table other files defer to | +| `knowledge/mailbox-kanban-model.md` | Executor arms, ahead-firing writer, kanbanstep trigger, budget | 2026-09-05 | ⊘ **PARTLY STALE** inline note (2026-09-05) on the structural-owner row: `KanbanActor`/`KanbanMsg` deleted 2026-08-05, file now read-only meta-awareness surface | +| `knowledge/compiled-templates.md` | The askama↔elixir-DSL analogy, DSL triple, oracle split, gap list | 2026-07-02 | "CORRECTED 2026-07-02" note retracting the original NextAction↔OgarAction 1:1 claim | +| `knowledge/d-mbx-a6-owner-consume-and-persistence.md` | D-MBX-A6 owner-consume adapter + fire-and-forget persistence sink spec | 2026-08-01 | none (itself documents the ack-elimination-compatible design) | +| `knowledge/multi-anchor-ast-resolution.md` | The ruff+odoo multi-anchor AST resolution method (harvest-adjacent) | 2026-07-02 | status line: "[G] for method existence, [H] per-mechanism until ground-truthed" | +| `knowledge/persona-vs-rung-ladder.md` | Demarcation: rung-content ladder (0–4) vs the persona-36 storyline; O1–O9 open items | 2026-08-30 | ⊘ SCOPE CORRECTION (2026-08-30, styles not confined to rung 4); several O-items internally regraded/re-scoped | +| `knowledge/sonnet-worker-guardrails.md` | §1 mandatory Sonnet-worker preamble + §2 vocabulary table + §3 footgun catalogue + §5 STOP triggers | 2026-08-04 | none itself; documents F1–F15 historical footguns | +| `soa_layout/README.md` | Doc map for the 4 soa_layout files | 2026-07-02 | none | +| `soa_layout/consumer-map.md` | 6-consumer write-path audit (T1–T4 tiers, warden table) | 2026-07-02 | none (content is a 2026-07-02 snapshot; not re-verified since) | +| `soa_layout/le-contract.md` | The 4+12 byte facet atom, L1–L8 payload catalogue, slot purity, §3b two-level LE contract | 2026-08-17 | §3a wide-carving waiting room explicitly a "grace period... not a tail revival"; multiple "open reconciliation items [H]" | +| `soa_layout/routing.md` | Address-as-router: prefix routing, mailbox routing, write routing, read-mode routing, adoption-as-range-count | 2026-07-02 | none | +| `soa_layout/tenants.md` | The 16 `ValueTenant` lanes with exact byte offsets (refreshed 2026-08-22/23) | 2026-08-23 | ⊘ 2026-07-28 refresh note: doc had drifted 4 tenants behind code; ⊘ 2026-07-28 TRANSIENT-READ CORRECTION on a since-resolved mismatch | +| `soa_layout/witness-nibble-lane.md` | The A9 `CausalWitnessFacet` sub-byte lane: why it is NOT a 9th §3 layout | 2026-07-29 | opens by naming its own citation error ("shipped citing `le-contract §3 L9 G24N4` — an entry that does not exist") | + +--- + +## Wave status W0–W6 + +Verdicts cross-checked against `.claude/board/STATUS_BOARD.md` (§V3 rows, +lines ~890–917) and, where STATUS_BOARD itself was stale relative to +INTEGRATION-PLAN's own Addendum-15 (2026-08-04) correction, that correction +is cited instead. + +### W0 — Ratify & document +- **Promise:** ship `.claude/v3/` tree + the V3 awareness layer (knowledge docs, 4 agent cards, `/v3` skill, `/v3-audit` command, CLAUDE.md+BOOT.md entrypoints). +- D-ids: D-V3-W0a, D-V3-W0b. +- Evidence (STATUS_BOARD:894–895): `"D-V3-W0a | ... | Shipped (this PR) | complete: 7/7 mappers synthesized; MODULE-TABLE = 304/304 files (21/21 census chunks); soa_layout 5/5 docs"`; `"D-V3-W0b | ... | Shipped (this PR) | 4 knowledge docs, 4 cards, skill+command registered"`. +- **Verdict: CLOSED.** The folder exists, all files present and readable; this session's own inventory confirms the tree. + +### W1 — Envelope & ownership (the keystone) +- **Promise:** `mailbox_owner()` stamp, ahead-firing batch writer + cast pairing, delegation cache, MailboxId minting, probes. +- D-ids: D-V3-W1a..e. +- Evidence (Addendum-15, INTEGRATION-PLAN.md:730–761, dated 2026-08-04): `"D-V3-W1b batch writer | SHIPPED — lance-graph-planner/src/batch_writer.rs (...). NOT 'new module' to be built. Zero production call sites — built-undriven"`; `"D-V3-W1c delegation cache | SHIPPED"`; `"D-V3-W1e probes | SHIPPED"`. Then a further, later correction (`knowledge/d-mbx-a6-owner-consume-and-persistence.md` + STATUS_BOARD D-MBX-A6-P3c/P3d/P4, dated 2026-08-01/08-02) shows the writer IS now driven: `"cycle_driver.rs:516 (cognitive_pass)"` is a real production caller, and D-MBX-A6-P4 (`STATUS_BOARD:1665`) shows a full cycle/WAL driver "**P4a–P4f Shipped (slice)**" with a "64k/17 falsifier" test. +- **Verdict: CLOSED for the mechanism, PARTIAL for fleet-wide adoption.** The keystone (cast pairing, delegation cache, ownership stamp, and — per the later P3c/P4 work not reflected in the wave table itself — a real driven caller and cycle/WAL closure) is built and tested. What remains open per the plan's own honesty ledger (STATUS_BOARD:1665): "actor-owned production wiring NOT proven ... cognitive-shader-driver/SoA thought NOT proven ... durability FAKE (contract-probe WalSink, until LanceShardSink)." + +### W2 — Kanban executors (two arms + structural owner) +- **Promise:** D-MBX-A6 adapter emit, per-mailbox board-as-TENANT, supervisor wiring, symbiont arm, 550ms budget, dispatch-speed probe. +- D-ids: D-MBX-A6, D-V3-W2a..e. +- Evidence: D-MBX-A6 arm #1 — STATUS_BOARD:1662 `"Shipped"` (StrategyOutcome carrier). **D-V3-W2a is the clearest open/contradictory item**: Addendum-12a (INTEGRATION-PLAN.md:486–510, 2026-07-02) specifies a NEW 10th `BoardAggregates` ValueTenant gated on T1–T6 tests + a batched classid mint. STATUS_BOARD:901 (undated re-read, presumably still 2026-07-10-era): `"Queued (GATED: Addendum-12a...)"`. But Addendum-15 (2026-08-04, INTEGRATION-PLAN.md:744) instead claims: `"D-V3-W2a kanban tenant | SHIPPED as ValueTenant::Kanban — the kanban×Rubicon per-node phase cursor in contract/canonical_node.rs:1622+ (8-byte tenant), per the one-mailbox-one-board ruling."` — but `soa_layout/tenants.md` (refreshed 2026-08-23) still lists `ValueTenant::Kanban` (tenant #9) as the **pre-existing per-row** tenant (`KanbanTenant`), the same one `mailbox-kanban-model.md` explicitly calls a *sibling*, never a substitute, for the per-mailbox board; no `BoardAggregates` tenant exists anywhere in the current 16-tenant table. **This looks like an unresolved contradiction, not a closure** — flagged below in Candidate epiphanies. + D-V3-W2b (supervisor wiring): STATUS_BOARD:902 `"Shipped"` (2026-07-10) but COMPONENT-MAP's own 2026-09-05 correction says the `KanbanActor` half was DELETED 2026-08-05 and "has NO assigned architectural responsibility... a session reading the W2b row as an open work item would build the exact thing the ruling struck" (`mailbox-kanban-model.md`, quoted verbatim above). +- **Verdict: PARTIAL / CONTRADICTORY.** Arm #1 (planner) shipped; the structural-owner design (ractor `KanbanActor`) was shipped, then architecturally superseded (`E-ACTOR-IS-NOT-THE-PHASE-PATH-1`, 2026-08-04) — real progress happened via `persist_sink::recover_and_apply` instead, but the wave table's own W2b row is stale and would mislead a session into re-building the retired actor path. W2a's real state is genuinely unclear from the docs as written (see above). + +### W3 — Compiled templates +- **Promise:** `StepMask` type, ElixirTemplate→graph-flow adapter, Rig oracle + compile-down loop, catalogue. +- D-ids: D-V3-W3a..d. +- Evidence: STATUS_BOARD:905 `"D-V3-W3a | ... | In PR (2026-07-10) | contract::step_mask::StepMask, +5 tests (866 lib green)"`; :906 `"D-V3-W3b | ... | Queued"`; :907 `"D-V3-W3c | ... | Queued"`; :908 `"D-V3-W3d | ... | Queued"`. `compiled-templates.md`'s own gap list (unchanged since 2026-07-02): `"StepMask type does not exist yet"` — contradicted by the shipped `contract::step_mask::StepMask` (the knowledge doc was never updated after W3a shipped). +- **Verdict: PARTIAL, mostly OPEN.** W3a (StepMask) shipped; W3b/c/d remain queued as of the latest STATUS_BOARD rows found. The controlling knowledge doc (`compiled-templates.md`) is stale on this point. + +### W4 — DTO ladder +- **Promise:** rename `ResonanceDto`→`PerturbationDto` (D-PERT-1), BusDto/cast pairing call sites, L4 learning-loop probe. +- D-ids: D-PERT-1, D-V3-W4a, D-V3-W4b. +- Evidence: STATUS_BOARD:951 `"D-PERT-1 | ... | Shipped (#630, 2026-07-02; verified in-code 2026-07-10 — row was stale)"`; :909 `"D-V3-W4a | ... | In PR (2026-07-10)"`; :910 `"D-V3-W4b | ... | Queued"`. +- **Verdict: PARTIAL.** D-PERT-1 CLOSED. W4a in-PR-as-of-last-record (likely shipped by now given W1's mainline progress, unverified here). W4b open. + +### W5 — Consumer adoption +- **Promise:** 9 sub-items (q2 re-bakes, cpic mereology, bake annotation, W5d probes, ladybug/smb-office pulls, smb-office migration, OGAR emit.rs fix, MedCare born-stamped, q2 dual-bake collapse). +- D-ids: D-V3-W5a..i. +- Evidence: STATUS_BOARD rows :911–917 are almost uniformly `"Queued"` (D-V3-W5a,b,c,e,f implied via consumer-map.md, g, h, i all cite "Queued" or reference the consumer-map §5 items, none marked Shipped). `consumer-map.md` §5 (2026-07-02 snapshot, not re-verified): explicitly names smb-office-rs's `LanceConnector::upsert` as the still-live ORPHAN-WRITE and the "first live migration" target — no evidence it has been migrated since. +- **Verdict: OPEN.** No W5 sub-item found Shipped in the material read; this is the least-advanced wave. + +### W6 — Monitor & retirement +- **Promise:** adoption/corpus scanner, `0x1000` marker retirement (P4, operator checkpoint), legacy alias retirement, custom-half opening. +- D-ids: D-V3-W6a, D-CCF-4, D-V3-W6b, D-V3-W6c. +- Evidence: STATUS_BOARD:915 `"D-V3-W6a | ... | In PR (counting logic shipped 9c55646 2026-07-02... runnable examples/adoption_scan.rs added 2026-07-10; Lance-dataset sweep = residue)"`. **D-CCF-4 (STATUS_BOARD:950) has itself been RESCINDED**: `"RESCINDED (operator 2026-07-03, E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1: v2/v3 coexist permanently by schema — retirement off the table)"` — directly contradicting the still-live `INTEGRATION-PLAN.md:110` row text (`"P4's trigger is DEFINED: adoption reads 100%"`) and `soa_layout/routing.md:106` (`"adoption 100% ⇒ P4 trigger (operator checkpoint) ⇒ marker deprecates"`), and the primer's own §5 framing (`v3-substrate-primer.md:103-105`, `"0x1000 ... temporary by declaration"`). None of README.md, INTEGRATION-PLAN.md, or the primer mention the rescission. +- **Verdict: OPEN, and the plan's own premise (0x1000 is temporary / P4 retires it) is STALE — superseded by an operator ruling the wave-plan text does not carry.** This is the single largest doc/board drift found in this task; flagged in Candidate epiphanies below. + +--- + +## COMPONENT-MAP counts + +`COMPONENT-MAP.md` uses six verdicts: REUSE / REPURPOSE / EXTEND / RETIRE / NEW / BLOCKED (plus one-off compounds: REUSE/EXTEND, RETIRE-toward-contract, BLOCKED→flip-on, CORRECTED). Tallying every row across §§1,3,4,5,6 (the subsystem sections proper; §2 is prose, §7 is the consumer table, counted separately below): + +| Verdict | Approx. count (rows, incl. compounds folded into nearest primary) | +|---|---| +| REUSE (incl. `REUSE (HW)`, `REUSE (disambiguate)`) | ~34 | +| EXTEND (incl. `REUSE/EXTEND`) | ~11 | +| REPURPOSE (incl. `REPURPOSE (rename)`) | ~6 | +| RETIRE (incl. `RETIRE-toward-contract`) | 4 | +| NEW | 3 | +| BLOCKED (incl. `BLOCKED/NEW`, `BLOCKED→flip-on`) | ~5–7 | +| CORRECTED (one-off) | 1 | + +The document's own cross-cutting summary (COMPONENT-MAP.md:136–139, verbatim, more authoritative than my re-tally): *"REUSE dominates — the ruled model is mostly wired-not-invented... The load-bearing NEW pieces are exactly three: batch writer + delegation cache (W1), board-as-tenant type (W2a), StepMask + control-flow closure (W3a/b)."* + +**§7 Consumers** (6 rows, own vocabulary): openproject-nexgen-rs REUSE; MedCare-rs "REUSE + **born-stamped gate (W5h)**"; OGAR "REUSE + **emit.rs 3× `as u16` post-flip mislabel (W5g)**"; q2 "REUSE/REPURPOSE"; woa-rs EXTEND; smb-office-rs BLOCKED. + +**Every RETIRE item, with retirement-PR status:** +1. `CollapseGateEmission` (contract, §1) — **retired, PR cited**: "already tombstoned (PR #477); comment-only remains." +2. `cognitive_stack.rs::ThinkingStyle` (12-space, §3) — RETIRE-toward-contract, **no single PR cited in COMPONENT-MAP itself**, but ENTROPY-MILESTONES M9 records the actual retirement mechanism: 5+3-council pass, commit `1a11038`, "FIVE divergent style tables killed" (2026-07-10), later partially REOPENED (2026-07-18) for 3 untouched 12-entry tables the sweep missed. +3. `bindspace.rs::BindSpace` (§6, "RETIRE (W7)") — **no PR; not yet retired.** Gate = "parity test (mailbox_soa.rs:1145) is the deletion gate"; `sonnet-worker-guardrails.md` §2 confirms current live status: "the legacy global store, MIGRATION IN PROGRESS → MailboxSoA... something to extend or to delete now [FORBIDDEN]... add new writers to it; remove it [FORBIDDEN]." +4. `ladybug-rs bind_space + CogRedis` (§6, "RETIRE (their repo)") — **no PR; external repo, gated on contract-pulls-only migration (W5e)**, itself unshipped per the W5 verdict above. + +So of 4 named RETIRE items, only 1 has landed with a cited PR; 1 landed via a different (council) mechanism; 2 remain fully open. + +--- + +## ENTROPY-MILESTONES + +26 rows (M1–M26). Rough status distribution as last recorded in the file itself (2026-07-23): + +- **SHIPPED / RESOLVED (mechanical gate green):** M5 (PR #477 tombstone), M6 (registry shipped, default-retirement open), M7 (resolved 2026-07-02 correction), M9 (resolved 2026-07-10, then partially reopened 2026-07-18), M15 (resolved 2026-07-02), M25 (shipped v1 2026-07-02), M26 (mechanism shipped 2026-07-23, semantic placement remains). +- **IN-FLIGHT:** M3 (stamp shipped, enforcement gated on W1 — likely more advanced now given W1's Addendum-15 progress, not re-verified against M3's own row), M4 (successor shipped, cutover pending W1), M12 (CycleBudget shipped 2026-07-10, load-balancer consumption open), M13 (doc-side only), M17 (doc corrected, code is W3 — W3 itself only partially shipped per the wave table above), M20 (residual resolved 2026-07-18), M24 (marked "PARTIAL 2026-07-17" — **this is now visibly stale**: STATUS_BOARD's D-MBX-A6-P3d/P4 rows (2026-08-01/08-02) show a full durable-witness + recovery + cycle/WAL closure with a "64k/17 falsifier" test, far beyond what M24's row text describes). +- **QUEUED / PENDING (no shipment recorded):** M8, M10, M11, M14, M16, M19, M21, M22, M23. +- **RULING-NEEDED (blocked on an operator decision):** M18 (sigma-chain Ω→Δ→Φ→Θ→Λ vs KanbanColumn 6-phase mapping — no ruling found in this session's reads). + +Net: roughly 7/26 cleanly shipped, ~6/26 in-flight (2 visibly under-reported relative to later board activity), 9/26 untouched, 1/26 blocked on a ruling that was never made. + +--- + +## FUTURE-DESIGN + VISION open items + +**FUTURE-DESIGN.md** (append-only, newest-first) — every item that reads as not-yet-done, quoted: + +- *"D-MTS-1 — Markov-as-stream parity on the DeepNSM corpus: the stream earns the Markov crown or it doesn't. Gates ALL VSA-path removal. The keystone. [next]"* (VISION.md §8, cross-referenced from FUTURE-DESIGN's migration arc) — **not run**. +- *"`StepMask` in contract (sibling of FieldMask)... W3a"* — shipped per STATUS_BOARD, but the D-V3-W3b/c/d successors it unblocks remain Queued. +- *"D-TTV-1 (engineering, envelope-auditor gated)"* — thinking tenants onto V3 substrate — open, gated on envelope-auditor sign-off never recorded as given. +- *"the load-balancing wiring (W2d) does not exist yet"* (COMPONENT-MAP.md:75, cited by FUTURE-DESIGN's addenda) — open. +- *"W2c symbiont arm: one dependency-uncomment + ~10 min cold build (BlockedColdBuild is deliberate); attempt in-container after 1–3, disk permitting."* — never confirmed attempted in the material read. +- *"D-EPIPHANY-SIG-1 (queued, [H]/CONJECTURE) — Hambly–Lyons epiphany-vs-rumination detector... Probe-gated"* — open, though note the 2026-09-01 literature harvest (see Harvest section below) substantially advances the Hambly-Lyons math this item depends on, **without FUTURE-DESIGN.md itself being updated to reflect it**. +- *"Style-as-class (D-TSC-2/3, after the batched mint)"* (VISION.md §4) — explicitly gated on a mint that per the W2a contradiction above may or may not have happened. +- *"[ASPIRATION, the honest crown]: thinking that compiles its own thinking"* (VISION.md §4) — named as aspiration, not claimed done. + +**VISION.md** — items explicitly tagged `[ASPIRATION]` or `[next]` (verbatim): + +- *"cannot stop thinking — active inference as the dispatch mechanism... [ASPIRATION-in-operation: the doctrine is designed and partially wired, not yet measured as a closed loop]"* (§0). +- *"whether the REAL awareness loop inherits this is precisely D-MTS-6b"* (§3) — the in-driver validation of the k*=1 comma-bit finding, not yet run. +- *"[ASPIRATION until D-MTS-6b]"* (§3, repeated fence on the "reconstructs more than it stores" claim). +- *"the endgame this ladder points at [ASPIRATION, the honest crown]: thinking that compiles its own thinking"* (§4). +- *"6. The compilation loop closes [ASPIRATION]: traces → templates → runbooks → families"* (§8, the final numbered road item). +- The whole §8 road (`D-MTS-1` → `D-TTV-1` → `D-MTS-6b` → `D-MTS-2/3` → `D-TSC-2/3` → compilation-loop-closes) is presented in **explicit dependency order**, and per the material read in this session, **none of D-MTS-1, D-TTV-1, or D-MTS-6b were found shipped** — meaning the entire VISION.md road is still at its first rung. + +--- + +## O1–O6 + +From `knowledge/persona-vs-rung-ladder.md` (current state, incorporating the file's own regrades): + +- **O1 — rung↔content wiring absent.** Still open. `RungLevel` carries Pearl causal-depth only; "No rung references `sigma_rosetta`, `verb_table`, `recipes`, or `StyleFamily`." +- **O2 — the orchestration→recipe edge does not exist.** Still open; **re-scoped 2026-08-30** from "rung-4→rung-3 edge" to "any-rung orchestration → `recipes::Recipe`" (per the O9 scope correction). `default_runbook()` still points at the persona vocabulary, not the recipe layer. +- **O3 — persona storyline unwired by design.** Still open, and intentionally so ("no rung/dispatch code should consume it as if it were reasoning"). +- **O4 — D-TSC-1 residue.** Still open: `PlannerStyleExt` not re-exported at `api::` (the q2 E0599 path); a mis-worded `#[deprecated]` note; two untethered lab-only name→ordinal tables (`wire.rs`, `auto_style.rs`). +- **O5 — probe results to ledger (run 2026-07-14).** Marked as **done as a probe run** (results recorded: p64-bridge `STYLES[ord%12]` dormant, `UNIFIED_STYLES` tethered-not-collapsed), but the two follow-on hardenings it names (`#[deprecated]` on `style_by_ordinal`) are not confirmed shipped in this session's reads. +- **O6 — triangle structure unbuilt.** Still open. "Rung 4's autopoiesis triangle (frozen × learned × exploration) has its three poles in separate subsystems... but no composed macro object." Landing zone named (`E-THINKING-STYLES-ARE-CLASSES-1`) but not built. + +(For completeness, not asked but present in the same file: O7 — two divergent rung-2 144-vocabularies, unresolved; O8 — a third off-canon 0–9 rung ladder in `learning::cognitive_frameworks::Rung`, unresolved; O9 — the rung-4 anchoring of StyleFamily itself regraded 2026-08-26 as a "scalar-era artifact" under the new tower/multi-stratum reading — this is the correction that re-scoped O2.) + +--- + +## Harvest — senses found + +Grepped `.claude/v3`, `.claude/knowledge`, `.claude/board/EPIPHANIES.md`, `.claude/board/LATEST_STATE.md`, 16 named `.claude/plans/*.md` files, plus (per the coordinator's added-scope message) `.claude/nexgen/` (harvest/ + plans/) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md`, all read in full where new. At least **five distinct senses** of "harvest" are in live use in this workspace: + +1. **The ruff `ruff_*_spo` AST/SPO harvest producer pattern** — the dominant sense workspace-wide. `ruff_cpp_spo`, `ruff_csharp_spo`, `ruff_python_spo`, `ruff_ruby_spo`, `ruff_sqlalchemy_spo`, `ruff_spo_triplet`, `ruff_spo_address` walk a source AST (C++/C#/Python/Ruby/SQLAlchemy) and emit `(subject,predicate,object)` triples / method-resolution manifests that feed OGAR mints and lance-graph-contract types. Evidence in `.claude/v3/knowledge/multi-anchor-ast-resolution.md:3`: *"READ BY: any session doing ruff-harvest / odoo-rs transcode / OGAR..."*; `.claude/v3/MODULE-TABLE.md:103`: *"harvested via ruff_python_dto_check+AST"*; `.claude/v3/FUTURE-DESIGN.md:39`: *"ruff_cpp_spo Stockfish harvest (oracle-only)"*. Also the huge `E-OCR-*` / tesseract-rs arc in `EPIPHANIES.md` (line 17117 etc.) — a byte-parity-proven consumer-side use of the same pattern. + +2. **Domain-specific harvest corpora produced by sense 1** — e.g. `aiwar-neo4j-harvest` (OSINT graph, `.claude/v3/MODULE-TABLE.md:135`, `unified-soa-rubikon-integration-v1.md:63`), `odoo_ontology.spo.ndjson` (`.claude/v3/MODULE-TABLE.md:103`), `r2il-harvest-pass1` (a ruff GitHub Release, `r2il-machine-semantic-contract-v1.md:1319`), `tesseract-rs/.claude/harvest/` (oracle/parity artifacts, confirmed present on disk), `woa-rs/vendor/ogit/v02-harvest/` (vendored OGIT TTL, confirmed present on disk). + +3. **Cross-repo pattern-transfer harvest** — the sense used pervasively in the older (pre-V3, 2026-05/06) consumer-integration plans: "harvesting" reusable code shapes (bridge templates, reconciler shells) from one consumer repo (MedCare-rs, smb-office-rs) into another (woa-rs). Heaviest concentration: `lance-graph-in-woa-rs-v1.md` (30 hits, e.g. line 356: *"woa-rs as integration target, harvesting XRechnung + parallelbetrieb"*), `lance-graph-business-logic-poc-via-woa-rs-v1.md` (10 hits), `soa-value-tenant-migration-v1.md` (a dedicated "harvest brief" plan format, 12 hits, e.g. line 1: *"# SoA Value-Tenant Migration — Plan v1 (harvest brief + 5+3 sign-off)"*). + +4. **"Idea/literature harvest" as a generic gathered-evidence ledger** — session-scoped sweeps that read prior art and file findings without shipping code. Two concrete instances read in full this task: + - `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` — a 5-Opus-auditor literature sweep (NARS/bilattice/EIG/Hambly-Lyons theorems) against shipped `epistemic_bassin.rs`/`ogar-loco` code, dated 2026-09-01, itself later **partially superseded 2026-09-02** by a same-file ⊘ notice (five of six "loco-core calls" it treated as constitutional were retracted; only `TERNLOG (0x86)` survives). + - `.claude/nexgen/harvest/` (11 files, 2026-09-05) — 4 read-only Sonnet "reader" reports (cascade/rolling-floor/HDR code census, Prozentrang/Shannon doctrine census, proprioception/EWA/Mexican-hat/mask-surface census, sigker/power-kernel census) + 7 Sonnet "sweeper" reports over 65 recent PRs (lance-graph #1126–#1175, OGAR #274–#298, ndarray #277–#301), feeding `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (a live, partially-shipped plan: D-NXG-1/4/5 shipped 2026-09-05 per `STATUS_BOARD.md:74-89`). + +5. **A same-word, different-referent collision worth flagging on its own**: the nexgen/literature-harvest material's **"v3"** (`.claude/nexgen/harvest/14-...:6`: *"24-axis basis v3 mirror (`epistemic_bassin::axes`)"*; `literature-harvest...md:39`: *"the 24-axis basis v3 is `0x0334`"*) is a **completely different "v3"** from `.claude/v3/`'s mailbox-kanban-facet substrate — it is version-3 of a now-**retracted** `ogar-epistemic`/`epistemic_bassin` axis design (`E-SIX-SEMANTIC-FAMILIES-MUST-NOT-IMPERSONATE-EACH-OTHER-1`, 2026-09-02). Similarly, `.claude/v3/`'s own text uses "nexgen" only to mean the unrelated **`openproject-nexgen-rs`** consumer repo (`COMPONENT-MAP.md:115,121`; `consumer-map.md:15,29`) — never `.claude/nexgen/`. Two independent naming collisions ("v3" and "nexgen") sit in the same workspace at the same time. + +--- + +## Harvest — v3 coverage + +**Sense 1 (ruff SPO harvest producer):** documented as INTEGRATED-BY-REFERENCE. `.claude/v3/knowledge/multi-anchor-ast-resolution.md` is entirely about this sense and is explicitly listed in `.claude/v3/README.md:42` doc map ("ruff/odoo transcode landings"). `MODULE-TABLE.md` cites two live harvest-fed modules (`odoo_ontology.rs`, `aiwar.rs`). **Coverage: YES, first-class.** + +**Sense 2 (domain harvest corpora):** referenced only incidentally, as evidence inside sense-1 rows (`aiwar-neo4j-harvest`, `odoo_ontology.spo.ndjson`). No dedicated inventory of harvest corpora exists in `.claude/v3/`. **Coverage: PARTIAL, by citation only.** + +**Sense 3 (cross-repo pattern-transfer harvest, the woa-rs/medcare-rs/smb-office-rs plans):** **NOT referenced anywhere in `.claude/v3/`.** These plans predate the V3 rulings (2026-05/06 vs 2026-07-02) and describe consumer-integration work orthogonal to the mailbox/kanban/facet substrate; `.claude/v3/soa_layout/consumer-map.md` DOES audit woa-rs/MedCare-rs/smb-office-rs write-paths (its own, V3-specific concern — write-on-behalf compliance), but never cites or incorporates the harvest-brief plans themselves. **Coverage: NO** (different concern, same consumer names, no cross-reference). + +**Sense 4a — `literature-harvest-2026-09-01-post-1132.md`:** **NOT referenced anywhere in `.claude/v3/`.** Grep for `"literature-harvest"` and `"2026-09-01"` across `.claude/v3` returns zero hits. The doc's own `**Scope:**` line frames it entirely in terms of the *nexgen* "v3" (`epistemic_bassin`/`ogar-loco` axes), never `.claude/v3/`'s facet/kanban substrate — genuinely a different subsystem, so the absence of cross-reference may be *correct* rather than a gap, modulo the naming collision noted above. + +**Sense 4b — `.claude/nexgen/harvest/` + `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md`:** **NOT referenced anywhere in `.claude/v3/`.** Grep for `"nexgen"`, `"NestedBands"`, `"E-NXG"`, `"D-NXG"` across `.claude/v3` returns zero hits except the unrelated `openproject-nexgen-rs` string (confirmed above). Conversely, the nexgen plan does touch V3-adjacent vocabulary — `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md:60`: *"The T1/T2 warden's NAMED test"* (referencing `.claude/knowledge/membrane-tiers.md`, a DIFFERENT membrane-tier doctrine, not the V3 agent cards); room 27 (line 145) explicitly connects to the Think-struct/VSA-trajectory doctrine from the top-level `CLAUDE.md`, and room 21 (line 139) cites the "256:256-by-classid ruling" — but never `.claude/v3/le-contract.md`'s L1–L8 catalogue, `tenants.md`'s tenant table, or the mailbox-kanban model, even though `NestedBands` is explicitly framed as "per (classid, version, idx)" — the exact V3 addressing space. **This is a genuine, unambiguous gap**: the nexgen harvest and plan operate directly on V3-shaped addresses (classid, version-keyed slabs, `WideFieldMask`/`FieldMask`) without citing or updating the V3 doc family that owns that addressing scheme. + +**Bottom line for the coordinator's question:** *No* — `.claude/v3/` does **not** currently incorporate either the nexgen harvest sweeps or the 2026-09-01 literature harvest, by any file:line evidence found. The nexgen harvest is the more consequential omission because its subject matter (mask-based row selection over classid/version-keyed slabs) is architecturally adjacent to — and in room 27 explicitly claims kinship with — the V3 substrate's own `Think.trajectory` and tenant-lane machinery, yet the two doc families do not reference each other in either direction. + +--- + +## Harvest — gaps to add + +If an "update v3 with the harvest" pass were undertaken, concretely: + +- **`.claude/v3/README.md` doc map** — add a row pointing at `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` for "row-mask/histogram/rank work over classid-addressed slabs," parallel to the existing `soa_layout/` row, since `NestedBands` is version-keyed exactly like `ValueTenant`/`ENVELOPE_LAYOUT_VERSION`. +- **`.claude/v3/soa_layout/tenants.md`** — the nexgen plan's room 27 (`nexgen-mask-histogram-thresholds-v1.md:145`) explicitly proposes reading `Think.trajectory` as `&NestedBands` (masks) instead of `Vsa16kF32`; if/when `D-NXG-*` lands a consumer wired into a real `ValueTenant`, `tenants.md` needs a new row (it currently has zero mask-histogram tenant) — file that should carry it: `tenants.md` §2. +- **`.claude/v3/knowledge/v3-substrate-primer.md` §6 ("what must NOT be reinvented")** — should gain a line for the newest, most sweeping V3-adjacent ruling found this session but absent from the primer: `E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (`.claude/board/EPIPHANIES.md:815`, 2026-09-05 operator ruling, verbatim: *"anything must be wired into SoA V3 substrate no exceptions except the causaledge64 adjacent as ALU legacy substrat"*) — this directly hardens and updates the primer's own §6 table (the `Vsa16kF32`/CausalEdge64 rows) but is not cited there. +- **`.claude/v3/soa_layout/witness-nibble-lane.md`** — the nexgen literature-harvest's Hambly-Lyons/Pillar-11 findings (D1/D2/D3/D9 in `literature-harvest-2026-09-01-post-1132.md`, esp. D3: *"a single `u8:u8` rail read as ONE scalar axis is d=1"*) are directly about the same sub-byte/rail-reading discipline `witness-nibble-lane.md` polices (its §3–4 "slot purity"/"loci not magnitudes" rules); no cross-reference exists in either direction. File that should carry it: `witness-nibble-lane.md` §5 (its own "Open — flagged as PROPOSALS" section) or a new companion note. +- **`.claude/v3/ENTROPY-MILESTONES.md`** — the nexgen plan's room 4 (`nexgen-mask-histogram-thresholds-v1.md:122`) proposes collapsing `Cascade::Band`, `FloorBand`, `QualityTracker` buckets, and OGAR's `residue_band` into one `NestedBands.bucket_index` — this is textbook entropy-milestone shape (N representations → 1 canonical), yet no M27 row exists for it. +- **`.claude/v3/COMPONENT-MAP.md` §6 (Shader/convergence/foundation)** — if D-NXG's room 27 (`Think.trajectory` as masks) ever ships, the existing `Vsa16kF32`-carrier rows in COMPONENT-MAP and the primer both need a REPURPOSE/RETIRE-candidate row added; today COMPONENT-MAP has no row for `NestedBands` at all. +- **Cross-link back, the other direction:** `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` itself never cites `le-contract.md`, `tenants.md`, or `routing.md` even though it is operating on the same classid/version keyspace those files own — a plan-side gap as much as a v3-side one. + +--- + +## Candidate epiphanies + +Facts a future session would need the "why" for, none of which are currently recorded as a dated board entry cross-referenced from `.claude/v3/`: + +1. **W2a's status is internally contradictory across the same plan file.** `INTEGRATION-PLAN.md` Addendum-12a (line 486–510, 2026-07-02) specifies a NEW gated `BoardAggregates` 10th tenant; Addendum-15 (line 744, 2026-08-04) instead claims W2a "SHIPPED as `ValueTenant::Kanban`" — the pre-existing per-row tenant #9, which `mailbox-kanban-model.md` itself calls a *sibling*, not a substitute, of the per-mailbox board. `soa_layout/tenants.md` (refreshed as recently as 2026-08-23) shows no `BoardAggregates` tenant in the current 16-lane table. Either Addendum-15 is wrong, or the design was quietly re-scoped without regrading Addendum-12a or `mailbox-kanban-model.md`'s "gap list" (which still lists "Per-mailbox kanban board carried as a TENANT" as an open gap, INTEGRATION-PLAN §"What the arms still need," `mailbox-kanban-model.md:123`). +2. **D-CCF-4 (the `0x1000` marker retirement) was RESCINDED 2026-07-03**, one day after the plan that names it as the P4 trigger was written — but `INTEGRATION-PLAN.md:110`, `soa_layout/routing.md:106`, and `knowledge/v3-substrate-primer.md:103-105` all still describe `0x1000` as "temporary" with a defined 100%-adoption retirement trigger, with no ⊘ notice anywhere in `.claude/v3/`. `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1` is the ruling; it is cited only in `STATUS_BOARD.md:950`, never mirrored into the V3 doc family. +3. **ENTROPY-MILESTONES M2 and M24 are stale relative to their own cross-referenced board rows.** M2 says "QUEUED" for the `PerturbationDto` rename; STATUS_BOARD (D-PERT-1) has said "Shipped (#630)" since 2026-07-02 — before M2's own file was last edited (2026-07-23). M24 says "PARTIAL 2026-07-17"; STATUS_BOARD's D-MBX-A6-P3d/P4 rows (2026-08-01/08-02, well before ENTROPY-MILESTONES' 2026-07-23 mtime is odd — actually P3d/P4 postdate M24's mtime, meaning the row was never updated after real progress landed) describe a materially more complete durable-witness/recovery/cycle-WAL system with a "64k/17 falsifier" test. Neither row was updated to match. +4. **`E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (2026-09-05)** is a sweeping "no exceptions" operator ruling directly on V3 substrate scope, landed the same day as `COMPONENT-MAP.md`'s last edit, yet is not mentioned in COMPONENT-MAP, the primer, or README — a session reading only `.claude/v3/` would not know this ruling exists. +5. **Two independent naming collisions share the string "v3" and "nexgen" respectively** across unrelated subsystems (see Harvest sense 5 above) — worth a one-line disambiguation note in `.claude/v3/README.md` given both are workspace-wide, high-frequency terms. + +--- + +## Uncertain + +- **MODULE-TABLE.md's 304-row census was not individually re-verified.** I read the header/lance-graph section in full (rows 1–128) and the tail of the ancestry-pipeline addendum (p64-bridge, cognitive-shader-driver, rows ~430–493), but the middle of the lance-graph-contract section (113 files) and the thinking-engine/other-crate ancestry rows were not exhaustively re-read row-by-row against current source — my confidence in the file's *overall shape and role* is high, but I cannot certify every individual duplication/tech-debt claim it makes is still current (the doc itself is dated 2026-07-10 and several of its own findings, e.g. the `UNIFIED_STYLES`/style-table-proliferation notes, were later partially resolved per `persona-vs-rung-ladder.md` O5 — I did cross-check that one, but did not attempt to verify the rest). +- **Whether D-V3-W4a, D-V3-W2d, D-V3-W3a etc. have shipped since their last "In PR" STATUS_BOARD entry (all dated 2026-07-10) was not independently re-verified against current source** — I relied on the two available cross-checks (STATUS_BOARD rows as last written, and INTEGRATION-PLAN's Addendum-15 self-correction for W1/W2 only) rather than grepping live crate source for `StepMask`, `CycleBudget`, etc. Given how much progress Addendum-15 revealed for W1/W2 alone, it is plausible several other "Queued"/"In PR" rows are also stale-in-the-optimistic-direction; this task's scope (V3 doc inventory) did not include a full source audit. +- **The exact current shape of `ValueTenant::Kanban` vs. any real per-mailbox board object** (Candidate epiphany 1) could not be resolved from documentation alone — it requires reading `crates/lance-graph-contract/src/canonical_node.rs` around line 1622+ and the current `MailboxSoaOwner`/`KanbanColumn` wiring, which was out of scope for a doc-only inventory. +- **The COMPONENT-MAP verdict tally in this report is an approximate hand-count**, not a mechanically generated one; compound verdicts (e.g. "REUSE/EXTEND") were folded by judgment call into one bucket each. The document's own qualitative summary (quoted verbatim) is the more reliable source for "REUSE dominates." diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md new file mode 100644 index 000000000..2679d0ae5 --- /dev/null +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md @@ -0,0 +1,266 @@ +# W-waves (SIMD) + jc pillars — verification inventory + +Repos as of session start: lance-graph `main` @ `aeebfb23`; ndarray branch +`claude/randomized-signature-projection` == `origin/master` tip `b9afcb9b`. +Read-only throughout; no cargo run, no edits, no commits. + +Docs read in full: `lance-graph/.claude/knowledge/ndarray-vertical-simd-alien-magic.md`; +`ndarray/.claude/knowledge/vertical-simd-consumer-contract.md` (no lance-graph-side +copy of the consumer-contract doc exists — confirmed absent by `ls`). + +--- + +## W1a primitives + +All 5 SHIPPED in `ndarray/src/`, across scalar (`simd_scalar.rs`), AVX-512 +(`simd_avx512.rs`), NEON (`simd_neon.rs`), and partially WASM (`simd_wasm.rs`, +i4-unpack + saturating_abs only — no gather/prefetch/popcnt there). No +dedicated AVX2-only file defines these types; `U16x8`/`I8x16`/`U64x8` are each +declared once per arch-file (scalar / avx512 / neon), so the "AVX2" backend in +the contract doc is served by the x86_64 file (`simd_avx512.rs`), not a +separate avx2 module. + +| # | TD id | Symbol | Status per TD entry (verbatim) | Found in code | Verdict | +|---|---|---|---|---|---| +| 1 | TD-NDARRAY-SIMD-UNPACK-I4-16D | `I8x16::from_i4_packed_u64`, `I8x16::lane_i8::`, `batch_packed_i4_16` | "Status: Open" (`TECH_DEBT.md:1762`) | `simd_scalar.rs:1684,1695,1971`; `simd_avx512.rs:2673,2694,3147`; `simd_neon.rs:2269,2283,2505`; `simd_wasm.rs:819,830` | **SHIPPED, STALE-DOC** (TD says Open, all 4 backends have it) | +| 2 | TD-NDARRAY-SIMD-SATURATING-ABS-I8 | `I8x16::saturating_abs`, `I8x32::saturating_abs` | "Status: Open" (`:1776`) | `simd_scalar.rs:1709,1737`; `simd_avx512.rs:2716,2937`; `simd_neon.rs:2301,2325`; `simd_wasm.rs:841` | **SHIPPED, STALE-DOC** | +| 3 | TD-NDARRAY-SIMD-GATHER | `U16x8::gather_u16`, `palette_lookup_u8x8` | "Status: Open" (`:1790`) | `simd_scalar.rs:1799,1863`; `simd_avx512.rs:2846,2900`; `simd_neon.rs:2356,2423` | **SHIPPED but PARTIAL — STALE-DOC.** The x86_64 ("avx512") impl at `simd_avx512.rs:2846` is, by its own doc comment (`:2832-2836`), "a scalar-loop polyfill (real AVX2 gather via `_mm256_i32gather_epi32` + downcast is tracked as a follow-up optimisation)" — the true-SIMD AVX2 gather the contract doc specified is NOT implemented; only the scalar-correctness-anchor fallback ships on every arch. | +| 4 | TD-NDARRAY-SIMD-PREFETCH | `prefetch_read_t0/t1/t2` | "Status: Open" (`:1803`) | `simd_scalar.rs:1887,1893,1899`; `simd_avx512.rs:3091,3104,3115`; `simd_neon.rs:2445,2462,2477` | **SHIPPED, STALE-DOC** | +| 5 | TD-NDARRAY-SIMD-POPCOUNT-U64 | `U64x8::popcnt`, `U64x8::xor_popcount`, `U64x4::popcnt` | "Status: Open" (`:1817`) | `simd_scalar.rs:1916,1934,1952`; `simd_avx512.rs:2987,3046`; `simd_avx2.rs:2153,2176,2195` (`U64x4`/parity) | **SHIPPED, STALE-DOC** | + +All five W1a `TECH_DEBT.md` entries carry `**Status:** Open` verbatim while +every symbol they specify exists in `ndarray/src`. None of the five entries +has been updated to reflect this — a systematic doc-lag, not an isolated slip. + +## W1b consumer migrations + +Raw-intrinsic census (pattern: `_mm512_|_mm256_|_mm_|core::arch::|vld1q|is_x86_feature_detected|is_aarch64_feature_detected`, +count = matching **lines**, current tree): + +| Crate / file | TD id | TD status (verbatim) | Raw-intrinsic lines found | Verdict | +|---|---|---|---|---| +| `crates/holograph/src/hamming.rs` | TD-SIMD-SWEEP-W1 | "Status: Open" | 25 | Open, accurate — not migrated | +| `crates/lance-graph/src/graph/blasgraph/types.rs` | TD-SIMD-SWEEP-W2 (half) | "Status: Open", cites "types.rs (22 raw ops)" | **0** | **STALE-DOC.** `types.rs` has been fully migrated: `hamming_distance_dispatch` (`types.rs:452-467`) now calls `ndarray::hpc::bitwise::hamming_distance_raw(a_bytes, b_bytes)` under `#[cfg(feature = "ndarray-hpc")]` with a scalar `count_ones` fallback otherwise. Zero raw intrinsics remain in this file. | +| `crates/lance-graph/src/graph/blasgraph/ndarray_bridge.rs` | TD-SIMD-SWEEP-W2 (half) | same entry, cites "ndarray_bridge.rs (60 raw ops)" | 66 | **Open, and still uses the exact violation the primitive was built to close** — `_mm512_popcnt_epi64` at lines 465 and 493, where `U64x8::popcnt`/`xor_popcount` (shipped, TD #5) should be consumed instead. | +| `crates/bgz17/src/simd.rs` | TD-SIMD-SWEEP-W3 (half) | "Status: Open" | 22 | Open, accurate. File still hand-rolls `SimdLevel` enum + `detect_simd()` (AP-SIMD-3/8) and both `_mm256_i32gather_epi32` (`:136`) and `_mm512_i32gather_epi32::<2>` (`:223`) directly — none of the shipped `gather_u16`/`palette_lookup_u8x8` primitives are consumed. | +| `crates/bgz17/src/prefetch.rs` | TD-SIMD-SWEEP-W3 (half) | "Status: Open" | 1 | Open, accurate (and slightly better than the TD text: the original aarch64 `_prefetch` call it also cited is already gone — replaced by a documented no-op citing `stdarch_aarch64_prefetch` instability, `prefetch.rs:98-108` — only the x86 `_mm_prefetch` at `:96` remains). | +| `crates/lance-graph-contract/src/mul.rs` (`pub mod i4_eval`, from `:727`) | TD-SIMD-SWEEP-W4 | "Status: Open" (P0) | 64 | Open, accurate — entire raw-intrinsic block (`_mm512_srli_epi64`, `_mm512_cmp*_epi64_mask`, `_mm512_loadu_si512`, etc., `:1086-1279`+) untouched by either W1a-#1 (`from_i4_packed_u64`) or W1a-#2 (`saturating_abs`), though both are shipped and ready to consume. | +| `crates/thinking-engine/src/engine.rs` | TD-SIMD-SWEEP-W5 | "Status: Open" (P3) | 1 (`is_x86_feature_detected!("avx512vnni")`, `:508`) | Open, accurate. The crate already imports `ndarray::simd_amx` (`:160`) and routes the actual matvec dispatch through `simd_amx::matvec_dispatch` (`:475`), but the standalone 3-tier feature-detect at `:503-510` still duplicates raw `is_x86_feature_detected!` rather than delegating tier selection to the polyfill. | + +**Totals:** 5 crates, 179 raw-intrinsic-bearing lines currently (holograph 25 + +lance-graph/blasgraph 66 + bgz17 23 + lance-graph-contract 64 + thinking-engine +1). The knowledge doc's cited "158-violation finding" (`E-SIMD-SWEEP-1`, +2026-05-16) is **stale as a live count** — files have grown since (mul.rs in +particular carries many more batch functions than PR #398's original scope). +The one genuine reduction since 2026-05-16 is `blasgraph/types.rs`, which went +from its cited 22 raw ops to 0 by routing through the already-shipped +`ndarray::hpc::bitwise::hamming_distance_raw` — but its sibling file in the +same TD entry, `ndarray_bridge.rs`, still carries all 60+ of its original +violations, so the entry as a whole is half-true, half-stale. + +**Net W1b verdict: 0 of 6 consumer-migration TD entries (W1–W5) are closed.** +All 5 primitives they depend on are shipped and ready; none of the 6 files +(types.rs excepted, already independently migrated by a different path) has +been switched over. + +## W1.5 items + +| # | TD id | Doc status (`ndarray-vertical-simd-alien-magic.md`) | TD-status (verbatim, `TECH_DEBT.md`) | Code reality | Verdict | +|---|---|---|---|---| +| 6 | TD-NDARRAY-SIMD-SIGNATURE-PDE-SWEEP | "SHIPPED" | heading itself says "(W1.5-#6, SHIPPED)"; body: "~~Deferred~~ **SHIPPED 2026-09-02.**" | `ndarray/src/hpc/signature_pde.rs:91` — `pub fn signature_pde_sweep(x: &[Vec], y: &[Vec]) -> f64` exists; `sigker/src/kernel.rs:35` imports it (per the doc; not independently re-verified against sigker source in this pass) | **CONFIRMED SHIPPED, TD entry consistent with code.** Lane type is `f64`/`Vec` (via internal `F64x8`), NOT the `F32x16` originally sketched — both docs already carry this correction. | +| 7 | TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION | Knowledge doc: "SHIPPED (ndarray PR #294)" — `randomized_signature_sweep`/`_sweep_with`/`_step` + `INCREMENT_EPSILON` | **`TECH_DEBT.md:1841` heading: "(W1.5-#7, DEFERRED)"; body `Status: Deferred`** | `ndarray/src/hpc/randomized_signature.rs:244,292` — `randomized_signature_sweep_with` and `pub fn randomized_signature_sweep(path: &[Vec], matrices: &[f64], biases: &[f64], state_dim: usize) -> Vec` both exist and compile-checkable by inspection | **CONFIRMED STALE-DOC — exactly the orchestrator's suspicion.** The primitive is shipped in ndarray; `TECH_DEBT.md`'s #7 entry was never updated off "Deferred" even though the knowledge doc (same repo, `.claude/knowledge/ndarray-vertical-simd-alien-magic.md:96-99`) already says "gate open... measured TRUE (2026-09-02)" and the sibling ndarray-side contract doc (`vertical-simd-consumer-contract.md:279-300`) carries a full "⊘ CORRECTED — SHIPPED (ndarray PR #294)" block. Consumer wiring (`sigker/src/randomized.rs:95` `RandomizedSignatureBuilder::encode`) is separately noted in that doc as "NOT yet wired... treat as in-flight" — not independently re-verified here (would require reading sigker source, out of this task's scope). | +| 8 | TD-NDARRAY-SIMD-LYNDON-PACK | Knowledge doc: gate open, "still unbuilt, but NO LONGER GATED" | `TECH_DEBT.md:1854` heading "(W1.5-#8, DEFERRED)"; body `Status: Deferred` | No `lyndon_pack`/`lyndon_unpack_batch`/`I16x16::lyndon_pack` symbol anywhere in `ndarray/src` (0 grep hits) | **CONSISTENT — genuinely open, not stale.** Both docs agree it's unbuilt; TD status "Deferred" matches (though the knowledge doc's own standing note says the gate is open, not blocked — "Deferred" as a TD status word is arguably imprecise but not factually wrong about implementation state). | + +**Cross-check against orchestrator's stated priors:** #6 and #7 (PR #293, #294) +ARE merged as claimed — confirmed directly against `ndarray/src/hpc/{signature_pde,randomized_signature}.rs`. #8's scalar prerequisite claim (`sigker::log_signature` exists via lance-graph PR #1150) was **not independently re-verified** in this pass (would require reading `crates/sigker/src/log_signature.rs`, which was out of the grep/read budget spent on the SIMD-primitive side) — flagged in Uncertain. The orchestrator's suspicion that TD's #7 entry is stale-DEFERRED is **CONFIRMED**. + +--- + +## Stale-doc findings + +1. **All five W1a `TECH_DEBT.md` entries say "Status: Open"** (`:1762,1776,1790,1803,1817`) while every symbol each specifies is shipped in `ndarray/src/{simd_scalar,simd_avx512,simd_neon}.rs` (line citations in the W1a table above). Not one W1a entry has been updated to "Shipped." +2. **TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (`TECH_DEBT.md:1841-1850`) says "DEFERRED"** while `ndarray/src/hpc/randomized_signature.rs:244,292` ships `randomized_signature_sweep_with`/`randomized_signature_sweep`, and the workspace's own knowledge doc (`ndarray-vertical-simd-alien-magic.md:96-99`) and the ndarray-side contract doc (`vertical-simd-consumer-contract.md:283-300`) both already record it SHIPPED as of ndarray PR #294. This is a genuine three-way inconsistency: two docs say shipped, one says deferred. +3. **TD-SIMD-SWEEP-W2 (`TECH_DEBT.md:1877-1884`) describes `blasgraph/types.rs` as carrying "22 raw ops"** at "types.rs:440-506" — the file at that path today has **zero** raw-intrinsic lines; `hamming_distance_dispatch` (`types.rs:452-467`) already delegates to `ndarray::hpc::bitwise::hamming_distance_raw`. The entry's other half (`ndarray_bridge.rs`, 66 raw lines including the exact `_mm512_popcnt_epi64` call the shipped `U64x8::popcnt` primitive was built to replace) is still accurate and still Open — so the entry is a stale compound: half done, filed as if neither half were. +4. **The "158-violation finding" cited by the knowledge doc (`ndarray-vertical-simd-alien-magic.md:123`, `E-SIMD-SWEEP-1`) is stale as a live number** — current census across the same 5 crates is 179 lines (see W1b table). This is drift from file growth, not from remediation; only 1 of 6 files (types.rs) shows a real reduction. + +--- + +## jc pillars table + +Source: `crates/jc/src/lib.rs` (read in full, 200 lines) + per-module grep for +`PillarResult::deferred` / `pub fn prove`. + +| # (as labeled in module doc, `lib.rs:1-29`) | Name | Module | Registered in `run_all_pillars` (`lib.rs:137-183`)? | `prove()` behavior | +|---|---|---|---|---| +| 1 | E-SUBSTRATE-1: bundle associativity @ d=10000 | `substrate.rs` | yes | executes (`substrate.rs:50`) | +| 2 | Cartan-Kuranishi: role_keys ≡ Cartan characters | `cartan.rs` | yes | **`PillarResult::deferred(...)` unconditionally** (`cartan.rs:15-16`, 22-line file total) | +| 3 | φ-Weyl: 144-verb collocation coverage | `weyl.rs` | yes | executes (`weyl.rs:35`) | +| 4 | γ+φ preconditioner: prolongation step reduction | `precond.rs` | yes | executes (`precond.rs:141`) | +| 5 | Jirak Berry-Esseen: weak-dep noise floor @ d=16384 | `jirak.rs` | yes | executes (`jirak.rs:128`) | +| 5b | Pearl 2³ mask-accuracy | `pearl.rs` | yes | executes (`pearl.rs:118`) | +| (no 6) | — | — | — | no module/label numbered "6" exists anywhere in `lib.rs`'s doc header or registry — the sequence explicitly jumps 5→5b→7 | +| 7 | Köstenberger-Stark: inductive mean on Hadamard 2×2 SPD | `koestenberger.rs` | yes | executes (`koestenberger.rs:248`) | +| 8 | Düker-Zoubouloglou: Hilbert-space CLT for AR(1) | `dueker_zoubouloglou.rs` | yes | executes (`dueker_zoubouloglou.rs:181`) | +| 9 | EWA-Sandwich: Σ-push-forward along multi-hop edge paths | `ewa_sandwich.rs` | yes | executes (`ewa_sandwich.rs:271`) | +| 9b | EWA-Sandwich 3D | `ewa_sandwich_3d.rs` | yes | executes (`ewa_sandwich_3d.rs:471`) | +| 10 | Pflug-Pichler: nested-distance Lipschitz on Sigma DN-trees | `pflug.rs` | yes | executes (`pflug.rs:290`) | +| 11 | Hambly-Lyons: signature uniqueness on tree-quotient | `hambly_lyons.rs` | yes | **feature-conditional**: `active::prove()` (`:575`, real math) under `--features hambly-lyons`; **`PillarResult::deferred(...)` by default** (`:737-745`, since `default = []` in `Cargo.toml:26` and `hambly-lyons = ["dep:sigker"]` is opt-in) | + +**"11/12 implemented, Pillar 2 deferred" claim: CONFIRMED as literally stated** +in `lib.rs:24-29` ("Pillars 1, 3, 4, 5, 5b, 7-11 are immediately executable +... Pillar 2 (Cartan-Kuranishi) remains deferred"). Counting the 12 vector +entries against that sentence: 11 named as executable (1,3,4,5,5b,7,8,9,9b,10,11 += 11 items) + 1 deferred (2) = 12. Arithmetic checks out. + +**One caveat the header doesn't spell out:** "immediately executable" for +Pillar 11 is true only when the crate is built `--features hambly-lyons`. +Under a **plain default-feature build** (`cargo run --example prove_it` with +no `--features`), `hambly_lyons::prove()` also returns a `deferred` result +(`hambly_lyons.rs:738-745`, same shape as `cartan::prove()`) — so a default +`prove_it` run shows **10 executing + 2 deferred**, not "11 executing + 1 +deferred," even though the code for Pillar 11 genuinely exists and is not a +stub (unlike Pillar 2, which has no non-deferred path at all — `cartan.rs` is +22 lines total, all comment + the deferred call). + +`solver_order.rs` (`crates/jc/src/solver_order.rs`) exists as an explicit +**NEW, separate slot** — its own module doc (`:1-10`) states it is +deliberately NOT "another Pillar 11" (two batteries already use that number +across repos: this crate's `hambly_lyons` = uniqueness, ndarray's +`hpc::pillar::signature` = kernel stability). It is **not** in the +`run_all_pillars()` vector (`lib.rs:137-183` has no `solver_order::prove` +entry) — so it is a 13th battery, outside the 12-pillar registry, run only via +its own example (`w3_battery_sweep.rs`) or directly. Its own `prove()` +(`solver_order.rs:226`) is also feature-gated identically to Pillar 11 +(`--features hambly-lyons`; deferred otherwise, `:299-306`). + +--- + +## jc W2–W5 legs + +**Numbering note (verified, not conflated with the SIMD W1a/W1b/W1.5 waves +above):** these W-numbers belong to a *different* plan, +`.claude/plans/pillar11-signature-certification-unification-v1.md` (read in +full). Its own waves are W0–W5; **W1 and W4 do NOT live in `crates/jc` at +all** — both are homed in `ndarray/crates/sigker-parity` (an excluded crate), +per the plan's §7 execution-record table and confirmed by an empty grep for +`\bW4\b` anywhere under `crates/jc` in lance-graph. Only W2, W3, W5 (and the +W6 "Theorem 5 lattice leg," an addition beyond the original 0-5 plan) have +in-tree jc artifacts. + +| Leg | Certifies | Declared status (plan `pillar11-signature-certification-unification-v1.md` §7, verbatim numbers) | Home | Gate / trigger, quoted | +|---|---|---|---|---| +| W2 | Depth-∞ uniqueness leg (M-2): the Goursat PDE kernel's deviation from the constant/identity path, both forward (tree-equivalence) and converse (non-tree loops) | "Shipped — converse law \|dev/area²−2\| = 0.0077; edge measured at area 2.5e-3, boundary at 2.5e-4" | `jc::hambly_lyons` (`hambly_lyons.rs:201-235`, `active::depth_infinity_leg`), pre-registered via `examples/w2_refinement_sweep.rs` + `examples/w2_area_edge.rs` | Pass conditions in code (`hambly_lyons.rs:616-619`): `pde.forward_max < PDE_FORWARD_EPS` (5.0e-5) `&& pde.area_law_err < PDE_AREA_LAW_TOL` (0.05) `&& pde.edge_dev > PDE_EDGE_MIN_DEV` (1.0e-5) `&& pde.below_edge_dev < PDE_FLOOR` (1.0e-6). Feature-gated `hambly-lyons`. | +| W3 | Solver-order advantage (M-1) + carrier fidelity (M-4): level-2-area-augmented Goursat coefficients vs increment-only, plus the area-domain-vs-kernel-scalar cancellation trap | "Shipped — 29.90x/41.71x advantage, silence exactly 0, kernel-scalar trap demonstrated" | **New pillar slot**, `jc::solver_order` (not "Pillar 11" — see above), `examples/w3_battery_sweep.rs` | Pass condition (`solver_order.rs:237`): `adv1 > 10.0 && adv2 > 10.0 && silence < 1e-12 && area_monotone && trap_fires`. Feature-gated `hambly-lyons`. | +| W5 | PowerSig scalability (M-5) — long-path Goursat solves | "**DEFERRED, trigger measured not fired** — fires at path length ~11585 (memory half); longest in-tree 4609" (plan §7); STATUS_BOARD row identical | No code module yet — only the trigger-check example, `examples/w5_trigger_check.rs` | **Quoted trigger criterion** (`w5_trigger_check.rs:3-7`): "the first real stream whose Goursat solve exceeds 1 GiB or 10 s." Constants in code: `const GIB: f64 = (1u64 << 30) as f64;` `const TIME_TRIGGER_S: f64 = 10.0;`. Memory half computed exactly: `mem_len = (GIB / 8.0).sqrt()` ≈ **11,585** (=√(2^27)). Time half is measured at runtime on the executing machine (not statically knowable from source) — the plan's §7 execution record reports it was measured at **~33,388** on a prior run. `longest_in_tree = jc::hambly_lyons::LONGEST_PATH_POINTS.max(jc::solver_order::LONGEST_PATH_POINTS)` = max(4609, 2049) = **4609** (both constants read directly: `hambly_lyons.rs:211` `pub const LONGEST_PATH_POINTS: usize = 3 * PER_SEG + 1` with `PER_SEG = 1536` → 4609; `solver_order.rs:54` `pub const LONGEST_PATH_POINTS: usize = M + 1` with `M = 2048` → 2049). **Gate state: OPEN (not fired)** — 4609 < 11,585, so even the cheaper (memory) half of the trigger has not been reached; not independently re-run in this pass (would require `cargo run --release ... --example w5_trigger_check`, out of scope for a read-only agent), value taken from the plan doc's own execution-record row, which is itself a prior session's measurement, not a live one from this pass. | +| W6 | "Theorem 5 lattice leg" — the finite-depth Hambly-Lyons certificate (Annals 171(1) 2010 §2.4 Theorem 5, exhaustive lattice-word check) | Not in the original 0-5 plan; appears only inside `hambly_lyons.rs`'s own module doc as an addition ("W6 leg", "the finite-depth certificate") | `jc::hambly_lyons::active::lattice_leg` (`hambly_lyons.rs:363-573`) | Pass condition (`:623-628`): `lat.reduced_merged == 0 && lat.treelike_max_dist < LATTICE_EPS (1e-12) && lat.depth2_false_merges >= 1 && lat.false_merge_unresolved == 0 && lat.false_merge_max_sep_depth <= hl_theorem2_depth(8) && lat.d1_classes == 7`. | + +**W1 and W4** (for completeness, not in-tree here): plan §7 records W1 "Shipped +— level-normalized err 3.518e-6, margin 28.4x" and W4 "Shipped — Cholesky over +64 paths + indefinite counterexample; concentration 0.0038 at N=1000," both +homed in `ndarray/crates/sigker-parity` — **not verified against ndarray +source in this pass** (out of the assigned scope, which was lance-graph's jc +crate + ndarray's SIMD primitives specifically). + +**W0** — hygiene wave (name-collision/misnomer/stale-note fixes across both +repos), plan §7: "Shipped (lance-graph #1111, ndarray #289)." + +--- + +## jc examples + +Full `[[example]]` list from `crates/jc/Cargo.toml` (read in full), with +`required-features`: + +| Example | `required-features` | +|---|---| +| `goursat_substrate_probe` | `["hambly-lyons"]` | +| `prove_it` | none | +| `sigma_probe` | none | +| `probe_p1` | none | +| `osint_edge_traversal` | none | +| `splat_to_ewa_bridge` | none | +| `splat_triangle_count` | none | +| `splat_lpa_label_propagation` | none | +| `splat_louvain_modularity` | none | +| `splat_jaccard_adamic_adar` | none | +| `splat_perturbationslernen` | none | +| `ontology_locality_probe` | none | +| `style_table_agreement` | none | +| `substrate_compare` | none | +| `w2_refinement_sweep` | `["hambly-lyons"]` | +| `w5_trigger_check` | `["hambly-lyons"]` | +| `w3_battery_sweep` | `["hambly-lyons"]` | +| `w2_area_edge` | `["hambly-lyons"]` | + +**Files present in `crates/jc/examples/` but with NO corresponding +`[[example]]` Cargo.toml entry** (Cargo auto-discovers `examples/*.rs` by +default since `autoexamples` was not seen disabled anywhere in the +`[package]` table read): `l9_loci_real_text.rs`, `partof_isa_vs_palette256.rs`, +`rung_divergence_reliability.rs`, `weather_substrate_reliability.rs`. These +build with default features (no `required-features` gate) since Cargo's +auto-discovery carries no such attribute; not independently confirmed to +compile in this pass (no cargo run performed). + +--- + +## Board rows + +`STATUS_BOARD.md` § `pillar11-signature-certification-unification-v1` +(`:205-215`) — matches the plan doc's §7 execution record **exactly**, +including the W5 "DEFERRED, trigger measured not fired... fires at path +length ~11585... longest in-tree 4609" line verbatim. No drift found between +board and plan for this section. + +Other `STATUS_BOARD.md` rows naming jc/Pillar/sigker (not all independently +verified — listed for completeness per the task's grep instruction): +- `D-OIF-4` — `ewa_sandwich_x8` in `crates/jc` over `ndarray::simd::F64x8`, status "Queued." +- `D-NXG-11` — "**Blocked** on Pillar-6 σ_step calibration" — note this + references "Pillar-6," which (per the jc pillars-table finding above) does + not exist as a jc-registry pillar; this is presumably a different + numbering scheme (a different plan's own pillar count, e.g. + `weather-substrate-poc-v2.md` or `dialectic-engine-v1.md`'s "six operator + pillars," line `:734`) — **not resolved in this pass**, flagged in + Uncertain. +- `D-MEP-0` — "jc Pillar-6/7 provers green in-checkout" — same "Pillar-6" + naming, status Queued. +- `TD-JC-CLIPPY-RED-ON-BASE-1` (`TECH_DEBT.md:50`) — heading itself says + "RESOLVED 2026-09-05 (lint sweep + `jc-proof.yml` clippy step, + operator-directed after #1181)" — consistent with `AGENT_LOG.md`'s + 2026-09-05 entry describing the same resolution. +- CI: `STATUS_BOARD.md:1160-1164`, `CI-JC` row — `.github/workflows/jc-proof.yml` + runs `prove_it` on every PR touching `crates/jc/` or `cam.rs`, "In PR" + status, 5-min timeout — not independently verified against the actual + workflow file in this pass. + +Plans referencing PowerSig/hambly/Pillar-11 beyond the pillar11 plan itself +(found via `grep -rl`, contents not read in this pass): +`thinking-engine-harvest-closure-v1.md`, `h268-probe-wave-v1.md`, +`dismech-causal-replay-v1.md`, `weather-substrate-poc-v2.md`, +`temporal-markov-and-style-classes-v1.md`. + +--- + +## Candidate epiphanies + +Facts established this pass, each with citations, suitable for the +orchestrator to promote into a board entry if it chooses: + +1. **All five W1a `TD-NDARRAY-SIMD-*` entries in `lance-graph/.claude/board/TECH_DEBT.md` are stale "Open" labels on shipped work.** Every primitive named in TD-NDARRAY-SIMD-UNPACK-I4-16D, -SATURATING-ABS-I8, -GATHER, -PREFETCH, -POPCOUNT-U64 exists in `ndarray/src/{simd_scalar,simd_avx512,simd_neon}.rs` (line numbers in the W1a table above). None of the five TD entries has had its Status line updated. +2. **TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (#7) is a three-way doc/TD inconsistency, confirmed.** ndarray's own knowledge doc and its consumer-contract doc both record it SHIPPED (ndarray PR #294, `randomized_signature_sweep`/`_sweep_with`/`_step` in `ndarray/src/hpc/randomized_signature.rs:244,292`), but `lance-graph/.claude/board/TECH_DEBT.md:1841-1850` still reads "Status: Deferred." This confirms the orchestrator's stated suspicion exactly. +3. **`gather_u16` (W1a-#3) ships everywhere as a scalar-loop correctness anchor, not as true SIMD gather.** Its own doc comment at `ndarray/src/simd_avx512.rs:2832-2836` states the real AVX2 `_mm256_i32gather_epi32` path "is tracked as a follow-up optimisation" — so the primitive satisfies the API contract (bounds-safe, three backends agree) but not the performance intent the W1a spec described. +4. **`blasgraph/types.rs` has already migrated off raw intrinsics independently of the tracked W1b wave**, via `hamming_distance_dispatch` → `ndarray::hpc::bitwise::hamming_distance_raw` (`types.rs:452-467`), while its sibling `ndarray_bridge.rs` (same TD-SIMD-SWEEP-W2 entry) still carries the exact `_mm512_popcnt_epi64` call the shipped `U64x8::popcnt` primitive exists to replace. The TD entry conflates the two files' states into one still-Open line. +5. **jc's "11/12 implemented" header claim is accurate as literally written, with one unstated caveat:** Pillar 11 (`hambly_lyons`) is only substantively executable under `--features hambly-lyons`; under the crate's own `default = []` feature set it returns the same `PillarResult::deferred(...)` shape as Pillar 2. A plain `cargo run --example prove_it` therefore shows 10 executing pillars, not 11, unless the feature flag is passed. +6. **The jc pillar numbering has no "Pillar 6"** — the module doc's own numbered list (`lib.rs:1-29`) runs 1,2,3,4,5,5b,7,8,9,9b,10,11, and the 12-entry `run_all_pillars()` vector matches that sequence exactly (no gap, no phantom entry). Any reference elsewhere in the board to "Pillar-6" (`D-NXG-11`, `D-MEP-0`) refers to a different plan's own pillar numbering, not the jc crate's registry — a naming collision across documents, unresolved in this pass. +7. **The pillar11-signature-certification-unification-v1 plan's W-numbering is disjoint from the SIMD wave's W1a/W1b/W1.5 numbering** — both use "W1"/"W2"/etc. as labels but for entirely different, unrelated work streams. A session searching "W2" or "W4" without first identifying which plan is meant will find real but unrelated matches (confirmed here: grepping `crates/jc` for `\bW4\b` returns nothing, because that wave lives in `ndarray/crates/sigker-parity`, not lance-graph). +8. **W5's PowerSig trigger is measurably far from firing.** `longest_in_tree` (4609, computed from the two exported `LONGEST_PATH_POINTS` constants) is well under a quarter of the memory-half trigger length (~11,585, computed exactly as `√(2^30/8)`), which is itself the *cheaper* of the two trigger halves per the plan's own prior measurement (time half ~33,388). W5 stays legitimately deferred by its own pre-registered, falsifiable criterion. + +--- + +## Uncertain + +- **Sigker source not independently read.** Every claim about `sigker::log_signature` existing (orchestrator's premise for #8's "prerequisite"), `sigker/src/kernel.rs:35` importing `signature_pde_sweep`, and `sigker/src/randomized.rs:95` still running its own scalar loop, is taken from the ndarray-side knowledge docs, not verified by reading `crates/sigker/src/*.rs` directly in this pass. If precision on sigker's own state is needed, a follow-up read of that crate is required. +- **W5's `time_len ≈ 33,388` figure is not a live measurement from this session** — it is quoted from the plan doc's own §7 execution-record row, itself the record of a prior autoattended run (2026-08-31). This agent did not run `cargo run --release ... --example w5_trigger_check` (read-only, no cargo). The memory-half figure (~11,585) IS independently re-derivable from the static constants in `w5_trigger_check.rs` without running anything, and was recomputed here rather than merely copied. +- **"Pillar-6" cross-reference** (`STATUS_BOARD.md` `D-NXG-11`, `D-MEP-0`) is unresolved — could not determine in this pass which document actually owns a "Pillar 6" (candidates not opened: `weather-substrate-poc-v2.md`, `dialectic-engine-v1.md`'s "six operator pillars"). +- **The four un-registered jc example files** (`l9_loci_real_text.rs`, `partof_isa_vs_palette256.rs`, `rung_divergence_reliability.rs`, `weather_substrate_reliability.rs`) were confirmed present by `ls` and confirmed absent from `Cargo.toml`'s `[[example]]` list by full read, but were not opened themselves — unknown whether they compile cleanly under default features, and unknown whether Cargo's implicit auto-discovery (rather than an explicit entry) is intentional or an oversight. +- **`neural-debug`, `lance-graph-planner`, and other workspace crates were not grepped for raw intrinsics** — the task scoped the raw-intrinsic census to the 5 crates named in the knowledge doc's per-workload table (holograph, lance-graph/blasgraph, bgz17, lance-graph-contract, thinking-engine); a wider sweep might turn up additional violations outside that named set. +- **The `is_x86_feature_detected!`/`is_aarch64_feature_detected!` grep pattern was included in the raw-intrinsic count** per the task's own symbol list; this inflates the "violation" count slightly relative to a strict "raw SIMD intrinsic call" definition (e.g., `bgz17/src/simd.rs`'s `detect_simd()` calls are architecturally a duplicate-dispatch anti-pattern, AP-SIMD-8, rather than a raw-instruction call per se — both are real violations per the workspace's own `simd-savant` doctrine, just different flavors). diff --git a/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md b/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md index b95ffec77..b45c3c8ea 100644 --- a/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md +++ b/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md @@ -11,6 +11,17 @@ invisible to the index's coverage column. That is a known gap, recorded on purpose: moving the file is a one-line change once the folder earns a second plan. Do not read the coverage column's silence as "untracked". +**V3 cross-link (added 2026-09-07, plan inventory).** `NestedBands` is keyed +`(classid, version, idx)` — the `.claude/v3/` keyspace (`soa_layout/routing.md`, +`le-contract.md`, `tenants.md`) — and room 27 below proposes it as the carrier +that `.claude/v3/knowledge/v3-substrate-primer.md` §6 row 3 demotes `Vsa16kF32` +from. Until 2026-09-07 neither folder cited the other; `.claude/v3/README.md` +(doc-map row), `.claude/v3/ENTROPY-MILESTONES.md` M27 and +`.claude/v3/FUTURE-DESIGN.md` (2026-09-07 block) now point here, and this +paragraph points back. Read past the name collision: "v3" in +`harvest/14-*` and the literature harvest means the retracted +`epistemic_bassin` 24-axis basis, not the V3 substrate. + --- ## 0. The one sentence diff --git a/.claude/plans/counterfactual-rung3-closure-v1.md b/.claude/plans/counterfactual-rung3-closure-v1.md index dc0519df9..efc2480c2 100644 --- a/.claude/plans/counterfactual-rung3-closure-v1.md +++ b/.claude/plans/counterfactual-rung3-closure-v1.md @@ -379,7 +379,10 @@ otherwise a green probe creates pressure to re-interpret the result rather than accept it. This is the anti-HARKing gate for D-CFR-3. **OD-3 — CascadeChannels8 tension (flagged, not resolved).** -`.claude/v3/FUTURE-DESIGN.md:101-105` names `layered.rs::CascadeChannels8` the +`.claude/v3/FUTURE-DESIGN.md` (§ *"thinking-engine carries unwired gems"*, the +`layered.rs::CascadeChannels8` bullet — the line-range citation that stood here +was replaced by this anchor 2026-09-07 after a FUTURE-DESIGN prepend moved it; +citation-decay gate) names `layered.rs::CascadeChannels8` the *"First wiring target — the confirmed NEXT edge"*, gate = M8 parity, and says it *"collapses into `causal_edge::CausalEdge64`'s signed mantissa slot"*. **That is the same mantissa slot M20 retires** (`ENTROPY-MILESTONES.md:45`). Wiring a new diff --git a/.claude/v3/COMPONENT-MAP.md b/.claude/v3/COMPONENT-MAP.md index c85a965f6..8a2c15c27 100644 --- a/.claude/v3/COMPONENT-MAP.md +++ b/.claude/v3/COMPONENT-MAP.md @@ -110,6 +110,7 @@ Meta/Plasticity widths differ persisted-vs-hot with no parity test; | p64-bridge | REUSE | pure stateless mapping, zero BindSpace/ownership deps — clean by design. **Note (2026-07-18):** `workspace.exclude` + zero `p64` dep — the convergence is a call-site wiring pattern (consumer imports both `p64` + `p64-bridge`, `p64-bridge/src/lib.rs:4`), NOT a buildable local crate dependency | | ndarray `MultiLaneColumn` | REUSE (HW) | the LE lane carrier; repurpose opportunity: back MailboxSoA's `Box<[u64]>` identity planes (64 B-aligned) | | ndarray hpc (fingerprint/cascade/blackboard/read_bgz7) | REUSE (HW) | hardware-tier; causal_diff.rs "self-reinforcement LoRA" flagged for a leak-check read | +| `planner::nested_bands::NestedBands` (D-NXG-1, shipped 2026-09-05, `E-NXG-21`) | EXTEND | the sealed, version-keyed row-mask histogram over `(classid, version)` — bucket i = `mask_ternlog::` by name, rank = bucket index. `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` room 27 proposes it as the carrier for `Think.trajectory` in place of the retired `Vsa16kF32` (primer §6 row 3); the five threshold vocabularies it collapses are ENTROPY M27. NOT a tenant yet (no `ValueTenant` row — `tenants.md` deliberately untouched); a REPURPOSE of the `Vsa16kF32` rows above becomes due only if room 27 lands. Row added 2026-09-07 (plan inventory: this map carried no reference to the nexgen work) | | ladybug-rs bind_space + CogRedis | RETIRE (their repo) | pre-V3; zero SoaEnvelope adoption confirmed; contract pulls only (W5e) | ## 7. Consumers (q2 / MedCare-rs / woa-rs / openproject-nexgen-rs / OGAR / smb-office-rs) diff --git a/.claude/v3/ENTROPY-MILESTONES.md b/.claude/v3/ENTROPY-MILESTONES.md index 10979960a..b49c313ce 100644 --- a/.claude/v3/ENTROPY-MILESTONES.md +++ b/.claude/v3/ENTROPY-MILESTONES.md @@ -19,7 +19,7 @@ | # | N representations | → Canonical survivor | Mechanical gate | Status | |---|---|---|---|---| | M1 | 2 classid split orders (canon-low legacy vs canon-high) | `CLASSID_ORDER = CanonHigh` + `_LEGACY` read aliases | corpus scan: zero old-form rows → alias retirement; adoption 100% → P4 (ONE two-metric range-count tool, W6a) | SHIPPED (#628 fleet); gate tool QUEUED | -| M2 | 2 concepts sharing the `ResonanceDto` name (mechanical Ψ field vs perspectival Three-Mountains) | Ψ → `PerturbationDto` (D-PERT-1); perspectival keeps the name | rename lands w/ deprecated alias; grep `dto::ResonanceDto` = alias-only; blast radius confirmed 7 files + 1 comment | QUEUED (small, mechanical) | +| M2 | 2 concepts sharing the `ResonanceDto` name (mechanical Ψ field vs perspectival Three-Mountains) | Ψ → `PerturbationDto` (D-PERT-1); perspectival keeps the name | rename lands w/ deprecated alias; grep `dto::ResonanceDto` = alias-only; blast radius confirmed 7 files + 1 comment | ~~QUEUED (small, mechanical)~~ **SHIPPED** (#630, 2026-07-02 — STATUS_BOARD `D-PERT-1` "verified in-code 2026-07-10"; this cell lagged the board by nine weeks, regraded 2026-09-07 plan inventory) | | M3 | implicit/undeclared write ownership | ONE stamp: `SoaEnvelope::mailbox_owner()` + cast pairing | `/v3-audit` check 4 = zero unstamped online writes; warden OWNED verdicts | stamp SHIPPED; enforcement gated on W1 batch writer | | M4 | singleton `BindSpace` sink + per-mailbox `MailboxSoA` coexisting | `MailboxSoA` per mailbox | parity test (mailbox_soa.rs:1145) green + engine_bridge cutover → **W7 deletes bindspace.rs**; grep `Arc` = zero | IN-FLIGHT (successor shipped; cutover pending W1; 2026-07-10 E-MARKOV-TEMPORAL-STREAM-1 sharpens the target: MailboxSoA + temporal stream + L4 palette tenants, NOT same-VSA-planes — see D-MTS-4) | | M5 | cross-mailbox carriers (Vsa16k-as-carrier, Baton, CollapseGateEmission, emit()) | zero-copy envelope; Lance columnar I/O = the only byte writer | resurrection grep (`/v3-audit` check 2) = tombstone comments only | SHIPPED (PR #477 tombstone) | @@ -46,9 +46,10 @@ | M21 | 3 dep-free hand-copies of the 16-byte NodeGuid LE encoder (q2 cpic, q2 fma/converge.rs, woa-rs erp/canon.rs — each "byte-identical", zero shared code) | one zero-dep `canon-node-bytes` extraction all three import | byte-parity test vs contract NodeGuid; grep local encoders = imports only | QUEUED (W5) | | M22 | 2 divergent q2 OSINT V3 bakes (crates/osint-bake canon-high 0x0700_0000 vs data/osint-v3 STALE pre-flip 0x1000_0700 dual-GUID scheme) | one canon-high bake against osint_classview.rs's 0x0700/0x0701 reservation | re-bake; grep pre-flip forms in q2 data/ = zero (or dual-alias-read only) | QUEUED (W5; latent until a reader assumes canon-high) | | M23 | 2 write-path doctrines coexisting (owner-stamped V3 writes vs smb-office-rs `LanceConnector::upsert` — the ONE live online consumer write, no stamp/classid/envelope) | all online consumer writes routed through the batch-writer cast | consumer-map §2 table shows zero ORPHAN-WRITE rows; warden green fleet-wide | QUEUED (W5 first live migration; medcare-soa writer BORN stamped as the prevention half) | -| M24 | 2 write-intent bookkeepers (batch-writer internal state vs kanban board AHEAD update) | ONE: the durable record is the Lance row's own LanceVersion — the writer stages ephemeral intent, Lance holds the truth (the "kanban board IS the WAL / ack = confirmation" framing is retired with the ack, see PARTIAL note) | recovery = a pinned-reference temporal.rs read of the row's version (`QueryReference::at` + deinterlace), NOT a replay from writer state; grep writer-internal id→version maps = zero; cast carries descriptor never bytes (zero-copy sink through NodeRowPacket); stacked-casts-never-refused test (melden macht frei — Addendum-7; sink coalesces naturally via live-store reads) | PARTIAL 2026-07-17 (E-ACK-ELIMINATED-1: the second bookkeeper's confirmation half is DELETED — batch_writer.rs is a pure intent recorder; "ack = confirmation" in this row is history — confirmation = the row's own LanceVersion via temporal read; remaining collapse = the intent board into SoA rows, gated on the W2a mint) | +| M24 | 2 write-intent bookkeepers (batch-writer internal state vs kanban board AHEAD update) | ONE: the durable record is the Lance row's own LanceVersion — the writer stages ephemeral intent, Lance holds the truth (the "kanban board IS the WAL / ack = confirmation" framing is retired with the ack, see PARTIAL note) | recovery = a pinned-reference temporal.rs read of the row's version (`QueryReference::at` + deinterlace), NOT a replay from writer state; grep writer-internal id→version maps = zero; cast carries descriptor never bytes (zero-copy sink through NodeRowPacket); stacked-casts-never-refused test (melden macht frei — Addendum-7; sink coalesces naturally via live-store reads) | PARTIAL 2026-07-17 (E-ACK-ELIMINATED-1: the second bookkeeper's confirmation half is DELETED — batch_writer.rs is a pure intent recorder; "ack = confirmation" in this row is history — confirmation = the row's own LanceVersion via temporal read; remaining collapse = the intent board into SoA rows, gated on the W2a mint) → **2026-09-07 regrade (plan inventory):** the collapse advanced past this cell — STATUS_BOARD `D-MBX-A6-P3d`/`P4` (2026-08-01/02) ship the durable witness + recovery + cycle/WAL closure with the 64k/17 falsifier; what remains is the W2a mint question itself, which INTEGRATION-PLAN states two ways (Addendum-12a: a NEW gated `BoardAggregates` tenant; Addendum-15: "SHIPPED as `ValueTenant::Kanban`", the pre-existing per-ROW tenant) — recorded on FUTURE-DESIGN 2026-09-07, not adjudicated here | | M25 | 2 persistence surfaces for orchestration state (graph-flow SessionStorage backend vs mailbox kanban board) | ONE: KanbanSessionStorage — Session transitions ARE KanbanMoves via the W1b writer; replay = rebuild from board | kill-mid-graph replay test: session resumes identically from board-recorded moves; grep non-kanban SessionStorage impls in lance-graph = zero | SHIPPED v1 2026-07-02 (rs-graph-llm graph-flow `kanban` feature: KanbanSessionStorage, snapshot+move-log; kill-mid-graph replay test GREEN — no repeats/gaps, pinned Rubicon column sequence; grep gate holds: no non-kanban SessionStorage impl in lance-graph) | | M26 | 2 belief representations (D-DIA reasoning arena `Belief`{CStmt+TruthValue, `belief.rs`} risks a NEW persistent belief store parallel to the M20 awareness carrier) | ONE: a `Belief` PERSISTS as a **reading across the M20 autopoiesis triangle** — statement→`awareness_facet::SpoFacet` semantic triple, witness→`SpoFacet` witness triple, truth→`cognitive_palette`/NARS; the in-memory arena stays the ephemeral working-set (data-flow cat-2), NO new SoA belief schema | a `Belief ↔ SpoFacet` mapping with a semantic-triple round-trip parity test (codebook-mediated: term-id→palette256² pair, gated on `PROBE-CODEBOOK-44`/M1); grep new belief-storage types in the SoA layer = zero | MECHANISM SHIPPED 2026-07-23 (`E-DIA-V4-FIELD-SEARCH-LOOP-1`): the `Belief↔SpoFacet` lossless byte round-trip is built + green (`planner/nars/facet_fold.rs`, `to_spo_facet`/`cstmt_from_spo_facet`, round-trip parity across all copulas incl. `Rel(65535)` + boundary term-ids) — a Belief IS a reading of the M20 awareness register, no new store. Remaining: the lossy codebook-mediated SEMANTIC placement (term→centroid, needs term embeddings). Was QUEUED (D-DIA; the codebook dependency PROBE-CODEBOOK-44 is now **mechanism-GREEN** 2026-07-23 `E-PROBE-CODEBOOK-44-MECHANISM-1` — the term→palette256² hierarchy is real, real-data ρ pending; remaining M26 work = the `Belief↔SpoFacet` round-trip mapping itself. The foveated field-search reads the persisted `SpoFacet`, not the arena — `E-FOVEATED-HHTL-TRIE-FIELD-SEARCH-1` fold correction) | +| M27 | ≥5 threshold/band vocabularies over ONE distance column: `ndarray::hpc::cascade::Band` (quarters of t), `perturbation-sim::FloorBand`, holograph `QualityTracker` buckets + `HdrCascade` `threshold_l0/l1/l2` / `DEFAULT_EXCITE/INHIBIT`, OGAR `residue_band`, and the `mu + kσ` reject floor (`.claude/nexgen/harvest/01-cascade-rolling-floor-hdr.md`) | ONE sealed `NestedBands` per `(classid, version)`: `bucket_index(row) = partition_point(masks, row)`; bucket i = `mask_ternlog::` BY NAME; Prozentrang = the index; rollover = a version-keyed mask split, never a rewrite (`.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` rooms 0–4, §2 T1/T2) | production callers of the five vocabularies = calls into `planner::nested_bands` only; PROBE-NXG-HIST-1/ROLL-1/FLOOR-1 stay green; no numeric threshold parameter on a T3 signature (room 19 reflective fence); `E-NXG-20` bounds it — the rank floor is the best ACHIEVABLE boundary, exact only without ties | mechanism SHIPPED (`crates/lance-graph-planner/src/nested_bands.rs`, D-NXG-1/4/5, `E-NXG-21`, 2026-09-05); the collapse of the callers is D-NXG-6/7/8 — QUEUED. Row added 2026-09-07 (plan inventory: this folder had no reference to the nexgen work) | ## The meta-rule (why this ledger exists) diff --git a/.claude/v3/FUTURE-DESIGN.md b/.claude/v3/FUTURE-DESIGN.md index 022bf6620..04422d036 100644 --- a/.claude/v3/FUTURE-DESIGN.md +++ b/.claude/v3/FUTURE-DESIGN.md @@ -11,6 +11,16 @@ --- +## 2026-09-07 — plan inventory pass: the harvest folders were invisible from here; four mirrors landed, one contradiction recorded + +| Ruling / design | One line | Wiring queue | +|---|---|---| +| `.claude/nexgen/` (2026-09-05 harvest + `plans/nexgen-mask-histogram-thresholds-v1.md`) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` | Neither cited `.claude/v3/`, and this folder cited neither — although `NestedBands` is keyed `(classid, version, idx)` (this keyspace) and room 27 proposes `Think.trajectory` as `&NestedBands` in place of the `Vsa16kF32` primer §6 row 3 demotes. Mirrors: README doc-map row + collision note, ENTROPY M27, this row; `tenants.md` gets a row only when a D-NXG consumer lands in a real `ValueTenant` (none yet — a row now would be a wish) | D-NXG-6/7/8 (the five-vocabulary collapse, M27); D-NXG-11 blocked on the "Pillar-6" σ_step placeholder — a jc-EXTERNAL numbering: jc's own registry has no Pillar 6 (1,2,3,4,5,5b,7…11), so the board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) name another document's pillars | +| `E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (operator 2026-09-05) | "anything must be wired into SoA V3 substrate no exceptions except the causaledge64 adjacent as ALU legacy substrat" — landed the day COMPONENT-MAP was last edited and was cited nowhere in this folder; now primer §6 last row + README ruling list | every BLOCKED/NEW row in COMPONENT-MAP is read through it | +| `D-CCF-4` RESCINDED (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) | STATUS_BOARD row carried the rescission for nine weeks while INTEGRATION-PLAN W6, `routing.md` §5 and primer §5 still described `0x1000` as temporary with a 100 %-adoption retirement trigger; ⊘ notes now at all three | W6b/W6c "post-P4" = post operator-checkpoint on the corpus proof, never post-retirement | +| W2a is stated two ways in one file — RECORDED, not adjudicated | INTEGRATION-PLAN Addendum-12a (2026-07-02): a NEW gated `BoardAggregates` tenant behind T1–T6 + a batched mint. Addendum-15 (2026-08-04): "SHIPPED as `ValueTenant::Kanban`" — the pre-existing per-ROW tenant #9 that `mailbox-kanban-model.md` calls a sibling, never a substitute; `tenants.md` (refreshed 2026-08-23) lists no `BoardAggregates`. Which is current needs `canonical_node.rs:1622+` and the owner wiring read, not a doc-only pass | owner-session / operator call; until then W2a reads PARTIAL, and M24's residue stays gated on it | +| `compiled-templates.md` gap list said `StepMask` "does not exist yet" | It shipped 2026-07-10 (`contract::step_mask::StepMask`, D-V3-W3a); the doc lagged eight weeks. Struck 2026-09-07 | W3b/c/d still Queued | + ## 2026-07-17 (sixth wave) — ACK ELIMINATED everywhere (operator directive) | Ruling | One line | Wiring queue | diff --git a/.claude/v3/INTEGRATION-PLAN.md b/.claude/v3/INTEGRATION-PLAN.md index d70ab866d..05be32a3f 100644 --- a/.claude/v3/INTEGRATION-PLAN.md +++ b/.claude/v3/INTEGRATION-PLAN.md @@ -111,6 +111,15 @@ Gate: replay equivalence green on every template change (template-smith rule). | D-V3-W6b | Legacy alias retirement (`CLASSID_*_LEGACY`, compat reader narrowing) | corpus proof = zero old-form rows; never before | | D-V3-W6c | Custom half opens: 64k ClassView render catalogue + template catalogue dispatch (completes W3d/F2) | post-P4 only | +> ⊘ **`D-CCF-4` RESCINDED (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`).** +> STATUS_BOARD's `D-CCF-4` row carried the rescission for nine weeks while this +> table, `soa_layout/routing.md` §5 and the primer §5 still described `0x1000` +> as temporary with a 100 %-adoption retirement trigger. Mirrored 2026-09-07 +> (plan inventory). The row stays as written (append-only); its trigger is +> void — v2/v3 coexist permanently by schema, `0x1000` remains a monitor, and +> "post-P4" on W6b/W6c means *post operator-checkpoint on the corpus proof*, +> never *post-retirement*. + --- ## Standing gates (every wave) diff --git a/.claude/v3/README.md b/.claude/v3/README.md index 276f48cb5..df0f60909 100644 --- a/.claude/v3/README.md +++ b/.claude/v3/README.md @@ -43,14 +43,27 @@ render/template catalogue. | waking an agent | `agents/BOOT.md` (the four `v3-*` cards + trigger routing) | | landing future design / post-V3 rulings | `FUTURE-DESIGN.md` (the meta board: ruling index, migration arc, thinking-engine gem wiring queue) | | keeping the vision (the WHY, graded) | `VISION.md` — the AGI-aspiring canon re-grounded 2026-07-10; every claim [G]/[RULING]/[ASPIRATION]; filigree-reviewed | +| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12; `NestedBands` is keyed `(classid, version, idx)` — THIS folder's keyspace — and its room 27 proposes it as the carrier primer §6 row 3 demotes `Vsa16kF32` from) + board `E-NXG-*` (2026-09-05) + `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02: of the loco calls it treated as constitutional only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory) — before that neither doc family cited the other | Shortcuts: `/v3` (bootload), `/v3-audit` (pre-commit conformance greps). Canonical ruling texts live on the board (`.claude/board/EPIPHANIES.md`, entries `E-V3-*`, `E-MAILBOX-KANBAN-NO-COLLAPSEGATE`, `E-COMPILED-THINKING-TEMPLATES`, `E-DTO-LADDER-OWNERSHIP-SPLIT`, `E-TWO-RESONANCES-SPLIT`, `E-RUFF-ODOO-MULTI-ANCHOR-AST`, -`E-V3-PLANNER-TWO-NATURES-AND-SPEED-PROBE`) — cite those in PRs, not -these mirrors. +`E-V3-PLANNER-TWO-NATURES-AND-SPEED-PROBE`, and — added to this list +2026-09-07 because it postdates every mirror in this folder — +`E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (operator 2026-09-05: +no exceptions to the V3 substrate; the one carve-out is CausalEdge64-adjacent +ALU legacy) and `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1` (operator +2026-07-03: `D-CCF-4` RESCINDED, `0x1000` never retires — see primer §5)) — +cite those in PRs, not these mirrors. + +**Two name collisions to read past (2026-09-07 plan inventory):** "v3" in +`.claude/nexgen/harvest/14-*` and in the literature harvest means version 3 +of the *retracted* `epistemic_bassin` 24-axis basis (`0x0334`), never this +folder; "nexgen" inside this folder (`COMPONENT-MAP.md` §7, +`soa_layout/consumer-map.md`) means the `openproject-nexgen-rs` consumer, +never `.claude/nexgen/`. ## The three load-bearing NEW pieces (everything else is wiring) diff --git a/.claude/v3/knowledge/compiled-templates.md b/.claude/v3/knowledge/compiled-templates.md index e4cc0e3c4..0dd0e7162 100644 --- a/.claude/v3/knowledge/compiled-templates.md +++ b/.claude/v3/knowledge/compiled-templates.md @@ -80,7 +80,7 @@ over the same canon concept, exactly like app render skins ## Gap list (see INTEGRATION-PLAN W3) -- `StepMask` type does not exist yet (mint in contract, sibling of FieldMask). +- ~~`StepMask` type does not exist yet (mint in contract, sibling of FieldMask).~~ ⊘ SHIPPED — `contract::step_mask::StepMask` (D-V3-W3a, STATUS_BOARD, 2026-07-10). This line lagged the board by eight weeks; corrected 2026-09-07 (plan inventory). The three gaps below stand (W3b/c/d Queued). - ElixirTemplate → graph-flow `GraphBuilder` adapter (rs-graph-llm side). - Rig oracle node + equivalence-gated compile-down loop (D-VCW-7 lineage). - Catalogue dispatch keyed by classid custom half (P4-gated). diff --git a/.claude/v3/knowledge/v3-substrate-primer.md b/.claude/v3/knowledge/v3-substrate-primer.md index f978ce457..a0cb9f5d8 100644 --- a/.claude/v3/knowledge/v3-substrate-primer.md +++ b/.claude/v3/knowledge/v3-substrate-primer.md @@ -103,6 +103,12 @@ Orchestration compiles like `askama ↔ ClassView × FieldMask`: - `0x1000` in the custom half = the **V3-adoption MONITOR**, temporary by declaration. P4's trigger is defined: adoption reads 100%; then the marker retires and the custom half opens for the 64k render/template catalogue. + > ⊘ **RESCINDED (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`; + > STATUS_BOARD `D-CCF-4` row has carried it since; mirrored here 2026-09-07, + > plan inventory).** v2/v3 coexist permanently by schema. The marker stays a + > MONITOR; nothing retires at 100 %; "P4" is an operator checkpoint on the + > corpus proof (W6a), not a retirement event. The bullet above is read + > through this note — append-only, not rewritten. - Canon-high is a **clustered index**: domain scans = key-range predicates; the adoption monitor + corpus-proof scanner are ONE two-metric range-count tool. @@ -116,6 +122,7 @@ Orchestration compiles like `askama ↔ ClassView × FieldMask`: | `Vsa16kF32` as cross-mailbox carrier | intra-compartment bundle math only (The Click, local) | | Ownership fields on `BusDto` | `SoaEnvelope::mailbox_owner()` + cast pairing | | Local classid bit math in consumers | contract/ogar-vocab composers | +| Anything wired OUTSIDE the SoA V3 substrate — a parallel store, a side map, a bespoke carrier, a second projection | the V3 substrate, **no exceptions**; the ONE carve-out is CausalEdge64-adjacent code as ALU legacy (`E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1`, operator 2026-09-05; row added 2026-09-07) | Cross-ref: `.claude/board/EPIPHANIES.md` (entries named above), `.claude/handovers/2026-07-02-classid-canon-high-flip-to-v3-thinking-sessions.md`, diff --git a/.claude/v3/soa_layout/routing.md b/.claude/v3/soa_layout/routing.md index 6ef2326c0..ae0e169bd 100644 --- a/.claude/v3/soa_layout/routing.md +++ b/.claude/v3/soa_layout/routing.md @@ -105,6 +105,12 @@ key-range counts over the same index — ONE two-metric scanner (W6a): (operator checkpoint) ⇒ marker deprecates ⇒ custom half opens for the render/template catalogue. + > ⊘ **"marker deprecates" is RESCINDED** (operator 2026-07-03, + > `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`; mirrored here 2026-09-07, plan + > inventory). Both range counts stay exactly as defined — the scanner is + > unaffected — but adoption 100 % is a checkpoint, not a retirement: v2/v3 + > coexist permanently by schema and `0x1000` remains a monitor. + Cross-ref: `le-contract.md` (bytes), `tenants.md` (what the value lanes mean), `consumer-map.md` (who writes what), primer §5, -board E-V3-MARKER-IS-A-MONITOR. +board E-V3-MARKER-IS-A-MONITOR (+ E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1). diff --git a/.claude/v3/soa_layout/witness-nibble-lane.md b/.claude/v3/soa_layout/witness-nibble-lane.md index bff9723ab..09e927b45 100644 --- a/.claude/v3/soa_layout/witness-nibble-lane.md +++ b/.claude/v3/soa_layout/witness-nibble-lane.md @@ -135,6 +135,15 @@ above. They may be ruled on independently of §§1–4, which stand on their own compares **bare** offsets (valid only for co-located rows); fabric-level `absolute_agreement` compares **absolute** events (`pos_a + o_a == pos_b + o_b`). One must become canonical and the other route through it. +- **P5** (added 2026-09-07, plan inventory — a cross-reference, not a new + proposal) — `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` + census row **D3** rules from the Hambly–Lyons side what §§3–4 rule from the + slot-purity side: *"a single `u8:u8` rail read as ONE scalar axis is d=1 and + carries only the endpoint"* (Diehl–Ebrahimi-Fard–Tapia Rem. 1.4), so the + canon "`u8:u8` is two bytes, never widened" is load-bearing for any + signature over a rail (`d ≥ 2` static precondition). Neither document + cited the other before this line. Whether D3 constrains **P1**'s choice of + primary axis is open. ## §6 Falsifier for this page From 162963a5fa003ef63943401cfa220791f55af5fb Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 7 Sep 2026 16:04:42 +0000 Subject: [PATCH 2/4] =?UTF-8?q?board:=20Codex=20review=20fixes=20on=20#121?= =?UTF-8?q?8=20=E2=80=94=20jc=20Pillar=206/7=20vs=209/9b=20is=20an=20inter?= =?UTF-8?q?nal=20numbering=20conflict;=20NestedBands=20(classid,=20version?= =?UTF-8?q?)=20key=20is=20PROPOSED,=20shipped=20struct=20is=20version-only?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Two P2 findings from the Codex review of #1218, both verified against the tree and corrected at every site the claim was mirrored: 1. jc EWA pair numbering. The inventory said "no Pillar 6 in jc / another document's numbering". Wrong: `crates/jc/src/ewa_sandwich.rs:1` labels itself `//! Pillar 6: Σ-Push-Forward as EWA-Sandwich` (registered at `lib.rs:168-170`) and `ewa_sandwich_3d.rs:687` says "Pillar 7", while the `lib.rs:16` header numbers the same pair 9 / 9b. It is a conflict INSIDE jc, not between jc and an external document. Corrected in PLAN-INVENTORY §5, §7 (new collision row) and §8. 2. NestedBands key. The inventory, E-SEVEN-HARVEST-SOURCES-…-1, the V3 README doc-map row, COMPONENT-MAP §6, ENTROPY M27, FUTURE-DESIGN row 1 and the nexgen plan cross-link all stated the sealed object is keyed `(classid, version)` as shipped. The shipped D-NXG-1 struct (`pub struct NestedBands`, `crates/lance-graph-planner/src/nested_bands.rs:91-105`) is version-keyed over ONE column with no classid; the classid key is the nexgen plan's PROPOSED room-26 shape. Every site now says so. Gates: append_only_gate origin/main OK (9 files, none shrank); citation_decay --since origin/main 0 new / 147 pre-existing (one anchor in the corrected E3 paragraph re-ordered so the true symbol sits adjacent to its citation — the backtick-pairing flip again); SUPERSESSION-INDEX regenerated last, byte-identical. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp --- .claude/board/EPIPHANIES.md | 2 +- .claude/board/PLAN-INVENTORY-2026-09-07.md | 31 +++++++++++++------ .../nexgen-mask-histogram-thresholds-v1.md | 10 ++++-- .claude/v3/COMPONENT-MAP.md | 2 +- .claude/v3/ENTROPY-MILESTONES.md | 2 +- .claude/v3/FUTURE-DESIGN.md | 2 +- .claude/v3/README.md | 2 +- 7 files changed, 34 insertions(+), 17 deletions(-) diff --git a/.claude/board/EPIPHANIES.md b/.claude/board/EPIPHANIES.md index fc67fe4a3..474c66de3 100644 --- a/.claude/board/EPIPHANIES.md +++ b/.claude/board/EPIPHANIES.md @@ -44,7 +44,7 @@ - **ternlogq** (`membrane-tiers.md`; OGAR #298; nexgen rooms 4/18/20): `TERNLOG 0x86` is the only surviving loco call and its value byte is the program; the eight named immediates are a complete mask ISA; room 18's family-separation rule is the mechanical "no joins". A crosswalk program = `RANK` + `TERNLOG` per hop. - **blasgraph** (`SYNERGY-MAP-S00-S07` §8.0; D-GTM-0j): integer-only HDR semiring algebra, no splat; the boundary with masks is TYPE — Boolean relations are masks, valued relations are semirings over CSR (D-GTM-0g unrun decides where). K0..K7, the SPO 2³ triadic projections, are the eight Boolean masks of one quad row. -**One object.** A sealed `NestedBands` + overlap matrix per `(classid, version)`, read by each source its own way — search, monitoring, rank, basin, drift, quorum, hop order, known-unknown census (nexgen room 26 said as much before the crosswalk reading existed). **The fence that keeps it one object and not a god-field:** rung, band, plasticity and "style as a byte string of immediates" are four different things (next entry). +**One object.** The shipped struct is version-keyed over one column with no classid: `pub struct NestedBands` at `crates/lance-graph-planner/src/nested_bands.rs:91-105`. A sealed NestedBands + overlap matrix per (classid, version) is the nexgen plan's PROPOSED room-26 shape, not the shipped D-NXG-1 key (Codex review on #1218 caught this entry's first draft implying the key was shipped) — read by each source its own way — search, monitoring, rank, basin, drift, quorum, hop order, known-unknown census (nexgen room 26 said as much before the crosswalk reading existed). **The fence that keeps it one object and not a god-field:** rung, band, plasticity and "style as a byte string of immediates" are four different things (next entry). ## 2026-09-07 — E-RUNG-BAND-AND-PLASTICITY-ARE-THREE-AXES-NEVER-ONE-LEVEL-FIELD-1 — level of processing, level of assertion, and how content changes do not share a byte diff --git a/.claude/board/PLAN-INVENTORY-2026-09-07.md b/.claude/board/PLAN-INVENTORY-2026-09-07.md index c60fb4a37..7a9321ebc 100644 --- a/.claude/board/PLAN-INVENTORY-2026-09-07.md +++ b/.claude/board/PLAN-INVENTORY-2026-09-07.md @@ -167,9 +167,13 @@ integrates sense 1 first-class (`knowledge/multi-anchor-ast-resolution.md`, README row) and cited **neither** `.claude/nexgen/` (2026-09-05, four reader reports + seven PR sweeps + the mask-histogram plan) **nor** `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` — although -`NestedBands` is keyed `(classid, version, idx)` (the V3 keyspace) and the -nexgen plan's room 27 proposes it as the carrier primer §6 demotes -`Vsa16kF32` from. Every "nexgen" string inside `.claude/v3/` meant the +the nexgen plan keys its sealed object `(classid, version)` — the V3 +keyspace — as its PROPOSED T2 shape (room 26), and its room 27 proposes it as +the carrier primer §6 demotes `Vsa16kF32` from. The shipped D-NXG-1 struct is +narrower: `pub struct NestedBands` (`crates/lance-graph-planner/src/nested_bands.rs:91-105`) +holds `version` + masks over ONE column and no classid, so V3 routing and class +isolation are not yet enforced by it (Codex review on #1218 caught this +document's first draft stating the proposed key as shipped; corrected). Every "nexgen" string inside `.claude/v3/` meant the `openproject-nexgen-rs` consumer. Landed in this PR: README doc-map row + collision note + two rulings added to @@ -210,9 +214,16 @@ jc: the 12-entry registry is literally "11/12 implemented, Pillar 2 deferred" (`lib.rs:24-29`) — with the unstated caveat that Pillar 11 (`hambly_lyons`) returns `PillarResult::deferred` under the crate's own `default = []` (`hambly_lyons.rs:737-745`; `Cargo.toml:26,30`), so a plain `prove_it` shows -**10 executing + 2 deferred**. There is **no Pillar 6** in jc (1,2,3,4,5,5b, -7,8,9,9b,10,11); the board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) name a -different document's pillars. `jc::solver_order` is a 13th battery outside +**10 executing + 2 deferred**. jc carries **two numberings for its EWA pair**: +the module docs label them Pillar 6 / Pillar 7 (`//! Pillar 6: Σ-Push-Forward as EWA-Sandwich` +at `crates/jc/src/ewa_sandwich.rs:1`; the 3D module asserts "Pillar 7" at +`ewa_sandwich_3d.rs:687`), while the `lib.rs` header numbers the same two +modules 9 / 9b (`9. EWA-sandwich Σ-push-forward` at `crates/jc/src/lib.rs:16`). +The board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) and +`mul-ewa-trust-propagation-v1` follow the module-doc numbering — they DO mean +jc's EWA sandwich, and the collision is internal to jc. (The W-waves tag-file +and this document's first draft said "no Pillar 6 in jc"; Codex review on +#1218 caught it.) `jc::solver_order` is a 13th battery outside `run_all_pillars()`. Four example files are present with no `[[example]]` entry (`l9_loci_real_text`, `partof_isa_vs_palette256`, `rung_divergence_reliability`, `weather_substrate_reliability`). W5's @@ -253,7 +264,7 @@ run, not re-run here). |---|---|---| | **v3** | `.claude/v3/` mailbox-kanban-facet substrate | version 3 of the *retracted* `epistemic_bassin` 24-axis basis (`0x0334`) in `nexgen/harvest/14-*` and the literature harvest | | **nexgen** | `.claude/nexgen/` (2026-09-05 harvest + plan) | `openproject-nexgen-rs` (consumer repo) — the only sense inside `.claude/v3/` until this PR | -| **Pillar 6 / Pillar 11** | jc registry: no Pillar 6; Pillar 11 = Hambly–Lyons uniqueness | board rows "Pillar-6/7" = another document's numbering; ndarray `hpc::pillar::signature` = a second "Pillar 11" (kernel stability) | +| **Pillar 6/7 / Pillar 11** | jc module docs: EWA-sandwich = Pillar 6, its 3D analogue = Pillar 7 (`ewa_sandwich.rs:1`, `ewa_sandwich_3d.rs:687`) — the numbering the board and `mul-ewa-trust-propagation-v1` use | jc `lib.rs` header: the same two modules are 9 / 9b — an INTERNAL jc conflict, not an external document; and ndarray `hpc::pillar::signature` is a second "Pillar 11" (kernel stability) beside jc's Hambly–Lyons uniqueness | | **W1 … W5** | SIMD waves W1a/W1b/W1.5 (`ndarray-vertical-simd-alien-magic.md`) | `pillar11-signature-certification-unification-v1` W0–W6 (W1/W4 live in `ndarray/crates/sigker-parity`) | | **harvest** | five senses (§4) | — | | **EWA** | three different operations share the word (lit-harvest #27; jc sandwich = rendering push-forward) | — | @@ -280,8 +291,10 @@ ride with it: masks win on Boolean relations at every density (D-GTM-0j — a TYPE boundary), 0 bytes/step (0k), chaining pays only while masks stay in L2 and survivors stay above 0.1 % (0n), and packed prefixes do NOT carry cross-domain relations (0l) — the FK columns do. The seven harvest sources -collapse onto one object: the version-keyed `NestedBands` (+ overlap matrix) -per `(classid, version)` — reveal-ahead / cache-key = Lance version; Shannon = +collapse onto one object: the version-keyed `NestedBands` (+ overlap matrix), +per `(classid, version)` in the plan's proposed shape (the shipped struct is +version-keyed over one column, no classid) — reveal-ahead / cache-key = Lance +version; Shannon = WHERE only and it lags the popcount budget; EWA ranks WHERE to look, never WHAT is true; a known unknown = a hop whose survivor mask has popcount > 1; Boolean → masks, valued → blasgraph semirings over CSR. And the fence: rung diff --git a/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md b/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md index b45c3c8ea..3a5e01ce4 100644 --- a/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md +++ b/.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md @@ -11,9 +11,13 @@ invisible to the index's coverage column. That is a known gap, recorded on purpose: moving the file is a one-line change once the folder earns a second plan. Do not read the coverage column's silence as "untracked". -**V3 cross-link (added 2026-09-07, plan inventory).** `NestedBands` is keyed -`(classid, version, idx)` — the `.claude/v3/` keyspace (`soa_layout/routing.md`, -`le-contract.md`, `tenants.md`) — and room 27 below proposes it as the carrier +**V3 cross-link (added 2026-09-07, plan inventory; corrected the same day after +Codex review on #1218).** This plan keys its sealed object `(classid, version)` +(room 26, PROPOSED) — the `.claude/v3/` keyspace (`soa_layout/routing.md`, +`le-contract.md`, `tenants.md`). The SHIPPED D-NXG-1 struct is narrower: +`pub struct NestedBands` (`crates/lance-graph-planner/src/nested_bands.rs:91-105`) +holds `version` + masks over one column and no classid, so V3 routing and class +isolation are NOT yet enforced by it. Room 27 below proposes it as the carrier that `.claude/v3/knowledge/v3-substrate-primer.md` §6 row 3 demotes `Vsa16kF32` from. Until 2026-09-07 neither folder cited the other; `.claude/v3/README.md` (doc-map row), `.claude/v3/ENTROPY-MILESTONES.md` M27 and diff --git a/.claude/v3/COMPONENT-MAP.md b/.claude/v3/COMPONENT-MAP.md index 8a2c15c27..079489331 100644 --- a/.claude/v3/COMPONENT-MAP.md +++ b/.claude/v3/COMPONENT-MAP.md @@ -110,7 +110,7 @@ Meta/Plasticity widths differ persisted-vs-hot with no parity test; | p64-bridge | REUSE | pure stateless mapping, zero BindSpace/ownership deps — clean by design. **Note (2026-07-18):** `workspace.exclude` + zero `p64` dep — the convergence is a call-site wiring pattern (consumer imports both `p64` + `p64-bridge`, `p64-bridge/src/lib.rs:4`), NOT a buildable local crate dependency | | ndarray `MultiLaneColumn` | REUSE (HW) | the LE lane carrier; repurpose opportunity: back MailboxSoA's `Box<[u64]>` identity planes (64 B-aligned) | | ndarray hpc (fingerprint/cascade/blackboard/read_bgz7) | REUSE (HW) | hardware-tier; causal_diff.rs "self-reinforcement LoRA" flagged for a leak-check read | -| `planner::nested_bands::NestedBands` (D-NXG-1, shipped 2026-09-05, `E-NXG-21`) | EXTEND | the sealed, version-keyed row-mask histogram over `(classid, version)` — bucket i = `mask_ternlog::` by name, rank = bucket index. `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` room 27 proposes it as the carrier for `Think.trajectory` in place of the retired `Vsa16kF32` (primer §6 row 3); the five threshold vocabularies it collapses are ENTROPY M27. NOT a tenant yet (no `ValueTenant` row — `tenants.md` deliberately untouched); a REPURPOSE of the `Vsa16kF32` rows above becomes due only if room 27 lands. Row added 2026-09-07 (plan inventory: this map carried no reference to the nexgen work) | +| `planner::nested_bands::NestedBands` (D-NXG-1, shipped 2026-09-05, `E-NXG-21`) | EXTEND | the sealed, version-keyed row-mask histogram — version-keyed over ONE column today (`pub struct NestedBands`, `crates/lance-graph-planner/src/nested_bands.rs:91-105`: `version` + masks, no classid); the `(classid, version)` keying is the plan's PROPOSED room-26 shape, so V3 routing / class isolation is not yet enforced by it — bucket i = `mask_ternlog::` by name, rank = bucket index. `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` room 27 proposes it as the carrier for `Think.trajectory` in place of the retired `Vsa16kF32` (primer §6 row 3); the five threshold vocabularies it collapses are ENTROPY M27. NOT a tenant yet (no `ValueTenant` row — `tenants.md` deliberately untouched); a REPURPOSE of the `Vsa16kF32` rows above becomes due only if room 27 lands. Row added 2026-09-07 (plan inventory: this map carried no reference to the nexgen work) | | ladybug-rs bind_space + CogRedis | RETIRE (their repo) | pre-V3; zero SoaEnvelope adoption confirmed; contract pulls only (W5e) | ## 7. Consumers (q2 / MedCare-rs / woa-rs / openproject-nexgen-rs / OGAR / smb-office-rs) diff --git a/.claude/v3/ENTROPY-MILESTONES.md b/.claude/v3/ENTROPY-MILESTONES.md index b49c313ce..49c8694cc 100644 --- a/.claude/v3/ENTROPY-MILESTONES.md +++ b/.claude/v3/ENTROPY-MILESTONES.md @@ -49,7 +49,7 @@ | M24 | 2 write-intent bookkeepers (batch-writer internal state vs kanban board AHEAD update) | ONE: the durable record is the Lance row's own LanceVersion — the writer stages ephemeral intent, Lance holds the truth (the "kanban board IS the WAL / ack = confirmation" framing is retired with the ack, see PARTIAL note) | recovery = a pinned-reference temporal.rs read of the row's version (`QueryReference::at` + deinterlace), NOT a replay from writer state; grep writer-internal id→version maps = zero; cast carries descriptor never bytes (zero-copy sink through NodeRowPacket); stacked-casts-never-refused test (melden macht frei — Addendum-7; sink coalesces naturally via live-store reads) | PARTIAL 2026-07-17 (E-ACK-ELIMINATED-1: the second bookkeeper's confirmation half is DELETED — batch_writer.rs is a pure intent recorder; "ack = confirmation" in this row is history — confirmation = the row's own LanceVersion via temporal read; remaining collapse = the intent board into SoA rows, gated on the W2a mint) → **2026-09-07 regrade (plan inventory):** the collapse advanced past this cell — STATUS_BOARD `D-MBX-A6-P3d`/`P4` (2026-08-01/02) ship the durable witness + recovery + cycle/WAL closure with the 64k/17 falsifier; what remains is the W2a mint question itself, which INTEGRATION-PLAN states two ways (Addendum-12a: a NEW gated `BoardAggregates` tenant; Addendum-15: "SHIPPED as `ValueTenant::Kanban`", the pre-existing per-ROW tenant) — recorded on FUTURE-DESIGN 2026-09-07, not adjudicated here | | M25 | 2 persistence surfaces for orchestration state (graph-flow SessionStorage backend vs mailbox kanban board) | ONE: KanbanSessionStorage — Session transitions ARE KanbanMoves via the W1b writer; replay = rebuild from board | kill-mid-graph replay test: session resumes identically from board-recorded moves; grep non-kanban SessionStorage impls in lance-graph = zero | SHIPPED v1 2026-07-02 (rs-graph-llm graph-flow `kanban` feature: KanbanSessionStorage, snapshot+move-log; kill-mid-graph replay test GREEN — no repeats/gaps, pinned Rubicon column sequence; grep gate holds: no non-kanban SessionStorage impl in lance-graph) | | M26 | 2 belief representations (D-DIA reasoning arena `Belief`{CStmt+TruthValue, `belief.rs`} risks a NEW persistent belief store parallel to the M20 awareness carrier) | ONE: a `Belief` PERSISTS as a **reading across the M20 autopoiesis triangle** — statement→`awareness_facet::SpoFacet` semantic triple, witness→`SpoFacet` witness triple, truth→`cognitive_palette`/NARS; the in-memory arena stays the ephemeral working-set (data-flow cat-2), NO new SoA belief schema | a `Belief ↔ SpoFacet` mapping with a semantic-triple round-trip parity test (codebook-mediated: term-id→palette256² pair, gated on `PROBE-CODEBOOK-44`/M1); grep new belief-storage types in the SoA layer = zero | MECHANISM SHIPPED 2026-07-23 (`E-DIA-V4-FIELD-SEARCH-LOOP-1`): the `Belief↔SpoFacet` lossless byte round-trip is built + green (`planner/nars/facet_fold.rs`, `to_spo_facet`/`cstmt_from_spo_facet`, round-trip parity across all copulas incl. `Rel(65535)` + boundary term-ids) — a Belief IS a reading of the M20 awareness register, no new store. Remaining: the lossy codebook-mediated SEMANTIC placement (term→centroid, needs term embeddings). Was QUEUED (D-DIA; the codebook dependency PROBE-CODEBOOK-44 is now **mechanism-GREEN** 2026-07-23 `E-PROBE-CODEBOOK-44-MECHANISM-1` — the term→palette256² hierarchy is real, real-data ρ pending; remaining M26 work = the `Belief↔SpoFacet` round-trip mapping itself. The foveated field-search reads the persisted `SpoFacet`, not the arena — `E-FOVEATED-HHTL-TRIE-FIELD-SEARCH-1` fold correction) | -| M27 | ≥5 threshold/band vocabularies over ONE distance column: `ndarray::hpc::cascade::Band` (quarters of t), `perturbation-sim::FloorBand`, holograph `QualityTracker` buckets + `HdrCascade` `threshold_l0/l1/l2` / `DEFAULT_EXCITE/INHIBIT`, OGAR `residue_band`, and the `mu + kσ` reject floor (`.claude/nexgen/harvest/01-cascade-rolling-floor-hdr.md`) | ONE sealed `NestedBands` per `(classid, version)`: `bucket_index(row) = partition_point(masks, row)`; bucket i = `mask_ternlog::` BY NAME; Prozentrang = the index; rollover = a version-keyed mask split, never a rewrite (`.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` rooms 0–4, §2 T1/T2) | production callers of the five vocabularies = calls into `planner::nested_bands` only; PROBE-NXG-HIST-1/ROLL-1/FLOOR-1 stay green; no numeric threshold parameter on a T3 signature (room 19 reflective fence); `E-NXG-20` bounds it — the rank floor is the best ACHIEVABLE boundary, exact only without ties | mechanism SHIPPED (`crates/lance-graph-planner/src/nested_bands.rs`, D-NXG-1/4/5, `E-NXG-21`, 2026-09-05); the collapse of the callers is D-NXG-6/7/8 — QUEUED. Row added 2026-09-07 (plan inventory: this folder had no reference to the nexgen work) | +| M27 | ≥5 threshold/band vocabularies over ONE distance column: `ndarray::hpc::cascade::Band` (quarters of t), `perturbation-sim::FloorBand`, holograph `QualityTracker` buckets + `HdrCascade` `threshold_l0/l1/l2` / `DEFAULT_EXCITE/INHIBIT`, OGAR `residue_band`, and the `mu + kσ` reject floor (`.claude/nexgen/harvest/01-cascade-rolling-floor-hdr.md`) | ONE sealed `NestedBands` per `(classid, version)` (the plan's TARGET key, room 26 — the shipped struct is version-keyed over one column, no classid): `bucket_index(row) = partition_point(masks, row)`; bucket i = `mask_ternlog::` BY NAME; Prozentrang = the index; rollover = a version-keyed mask split, never a rewrite (`.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` rooms 0–4, §2 T1/T2) | production callers of the five vocabularies = calls into `planner::nested_bands` only; PROBE-NXG-HIST-1/ROLL-1/FLOOR-1 stay green; no numeric threshold parameter on a T3 signature (room 19 reflective fence); `E-NXG-20` bounds it — the rank floor is the best ACHIEVABLE boundary, exact only without ties | mechanism SHIPPED (`crates/lance-graph-planner/src/nested_bands.rs`, D-NXG-1/4/5, `E-NXG-21`, 2026-09-05); the collapse of the callers is D-NXG-6/7/8 — QUEUED. Row added 2026-09-07 (plan inventory: this folder had no reference to the nexgen work) | ## The meta-rule (why this ledger exists) diff --git a/.claude/v3/FUTURE-DESIGN.md b/.claude/v3/FUTURE-DESIGN.md index 04422d036..00f6a556d 100644 --- a/.claude/v3/FUTURE-DESIGN.md +++ b/.claude/v3/FUTURE-DESIGN.md @@ -15,7 +15,7 @@ | Ruling / design | One line | Wiring queue | |---|---|---| -| `.claude/nexgen/` (2026-09-05 harvest + `plans/nexgen-mask-histogram-thresholds-v1.md`) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` | Neither cited `.claude/v3/`, and this folder cited neither — although `NestedBands` is keyed `(classid, version, idx)` (this keyspace) and room 27 proposes `Think.trajectory` as `&NestedBands` in place of the `Vsa16kF32` primer §6 row 3 demotes. Mirrors: README doc-map row + collision note, ENTROPY M27, this row; `tenants.md` gets a row only when a D-NXG consumer lands in a real `ValueTenant` (none yet — a row now would be a wish) | D-NXG-6/7/8 (the five-vocabulary collapse, M27); D-NXG-11 blocked on the "Pillar-6" σ_step placeholder — a jc-EXTERNAL numbering: jc's own registry has no Pillar 6 (1,2,3,4,5,5b,7…11), so the board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) name another document's pillars | +| `.claude/nexgen/` (2026-09-05 harvest + `plans/nexgen-mask-histogram-thresholds-v1.md`) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` | Neither cited `.claude/v3/`, and this folder cited neither — although the nexgen plan keys its sealed object `(classid, version)` (this keyspace) as its PROPOSED room-26 shape and room 27 proposes `Think.trajectory` as `&NestedBands` in place of the `Vsa16kF32` primer §6 row 3 demotes. The SHIPPED D-NXG-1 struct is version-keyed over one column with no classid (`pub struct NestedBands`, `crates/lance-graph-planner/src/nested_bands.rs:91-105`), so V3 routing / class isolation is not yet enforced by it (Codex on #1218 caught the first draft of this row saying otherwise). Mirrors: README doc-map row + collision note, ENTROPY M27, this row; `tenants.md` gets a row only when a D-NXG consumer lands in a real `ValueTenant` (none yet — a row now would be a wish) | D-NXG-6/7/8 (the five-vocabulary collapse, M27); D-NXG-11 blocked on the "Pillar-6" σ_step placeholder — jc's OWN EWA-sandwich pillar: its module doc labels it Pillar 6 (`crates/jc/src/ewa_sandwich.rs:1`) while the `lib.rs` header numbers the same module 9 — an internal jc numbering conflict, not an external document (first draft of this row said the opposite; Codex on #1218) | | `E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (operator 2026-09-05) | "anything must be wired into SoA V3 substrate no exceptions except the causaledge64 adjacent as ALU legacy substrat" — landed the day COMPONENT-MAP was last edited and was cited nowhere in this folder; now primer §6 last row + README ruling list | every BLOCKED/NEW row in COMPONENT-MAP is read through it | | `D-CCF-4` RESCINDED (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) | STATUS_BOARD row carried the rescission for nine weeks while INTEGRATION-PLAN W6, `routing.md` §5 and primer §5 still described `0x1000` as temporary with a 100 %-adoption retirement trigger; ⊘ notes now at all three | W6b/W6c "post-P4" = post operator-checkpoint on the corpus proof, never post-retirement | | W2a is stated two ways in one file — RECORDED, not adjudicated | INTEGRATION-PLAN Addendum-12a (2026-07-02): a NEW gated `BoardAggregates` tenant behind T1–T6 + a batched mint. Addendum-15 (2026-08-04): "SHIPPED as `ValueTenant::Kanban`" — the pre-existing per-ROW tenant #9 that `mailbox-kanban-model.md` calls a sibling, never a substitute; `tenants.md` (refreshed 2026-08-23) lists no `BoardAggregates`. Which is current needs `canonical_node.rs:1622+` and the owner wiring read, not a doc-only pass | owner-session / operator call; until then W2a reads PARTIAL, and M24's residue stays gated on it | diff --git a/.claude/v3/README.md b/.claude/v3/README.md index df0f60909..02e696061 100644 --- a/.claude/v3/README.md +++ b/.claude/v3/README.md @@ -43,7 +43,7 @@ render/template catalogue. | waking an agent | `agents/BOOT.md` (the four `v3-*` cards + trigger routing) | | landing future design / post-V3 rulings | `FUTURE-DESIGN.md` (the meta board: ruling index, migration arc, thinking-engine gem wiring queue) | | keeping the vision (the WHY, graded) | `VISION.md` — the AGI-aspiring canon re-grounded 2026-07-10; every claim [G]/[RULING]/[ASPIRATION]; filigree-reviewed | -| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12; `NestedBands` is keyed `(classid, version, idx)` — THIS folder's keyspace — and its room 27 proposes it as the carrier primer §6 row 3 demotes `Vsa16kF32` from) + board `E-NXG-*` (2026-09-05) + `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02: of the loco calls it treated as constitutional only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory) — before that neither doc family cited the other | +| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12; the plan's sealed object is keyed `(classid, version)` — THIS folder's keyspace — as its PROPOSED room-26 shape, while the shipped D-NXG-1 `NestedBands` is version-keyed over one column with no classid; its room 27 proposes it as the carrier primer §6 row 3 demotes `Vsa16kF32` from) + board `E-NXG-*` (2026-09-05) + `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02: of the loco calls it treated as constitutional only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory) — before that neither doc family cited the other | Shortcuts: `/v3` (bootload), `/v3-audit` (pre-commit conformance greps). Canonical ruling texts live on the board (`.claude/board/EPIPHANIES.md`, From f5d530bb1ba4e4450c8dbd0ab503bbe9c0969dab Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 7 Sep 2026 16:17:29 +0000 Subject: [PATCH 3/4] =?UTF-8?q?board:=20CodeRabbit=20review=20fixes=20on?= =?UTF-8?q?=20#1218=20=E2=80=94=20GATHER=20is=20PARTIAL=20not=20SHIPPED;?= =?UTF-8?q?=20W1b=20is=200/5=20TD=20entries=20+=201/7=20files;=20M1?= =?UTF-8?q?=E2=80=93M27=20reconciled;=20StepMask=20regraded=20in=20COMPONE?= =?UTF-8?q?NT-MAP=20+=20guardrails;=200x1000=20permanent=20in=20README=20+?= =?UTF-8?q?=20guardrails;=20tag-file=20errata?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit CodeRabbit posted 12 findings against ddd3393b. Each was verified against the tree; the substantive ones are corrected in the consolidated documents, the tag-file ones are answered by an append-only "Orchestrator errata" block at the end of each affected tag-file (the agent text stays verbatim — it is the evidence the inventory cites). Substantive (consolidated docs): - TD-NDARRAY-SIMD-GATHER: leading token PARTIAL, not SHIPPED. The API exists on scalar/avx512/neon but the x86 body is a scalar-loop polyfill by its own doc comment (`simd_avx512.rs:2833-2836`); the AVX2 gather the row requires is still open. Inventory §0/§1a/§5 and the EPIPHANIES numbers line now say 4/5 shipped + GATHER partial. - W1b: "0 of 6" mixed TD entries with files. There are 5 TD entries (TD-SIMD-SWEEP-W1..W5, 0 closed, W2 half-done) over 7 files (1 migrated — `types.rs` — 6 still raw). Inventory §0/§5 + EPIPHANIES numbers line. - ENTROPY-MILESTONES: the §3 breakdown omitted M1 (SHIPPED #628, gate tool queued) and §0 summed to 24. Now 9 shipped + 7 in-flight + 10 queued + 1 ruling-needed = 27. - STATUS_BOARD buckets: 575 of 592 rows classified; the 17 outside the nine leading tokens are now stated as unclassified (§0/§6), not implied away. - StepMask: COMPONENT-MAP still said BLOCKED/NEW "zero .rs matches" and the Sonnet guardrails vocabulary said "does not exist yet"; both regraded to `pub struct StepMask` (`crates/lance-graph-contract/src/step_mask.rs:40`, exported `lib.rs:241`, D-V3-W3a 2026-07-10). - 0x1000: the README five-sentence summary and the guardrails vocabulary row still called it temporary with a P4 retirement; both now carry the D-CCF-4 rescission (permanent monitor; P4 = operator checkpoint). - README doc-map row for the nexgen plan rewritten into complete clauses. - EPIPHANIES E-PLUG-AND-PLAY-…-1 Status line: "Fixed in this PR" was a context line from #1216 (commit 296e8b79), not this PR's change; it now names #1216 so the phrase does not go stale after merge (Status-line edits are the one permitted edit on a past entry). Tag-file errata (append-only, one block per file): the two truncated status_token cells with the full source values; the three concatenated headings (left as written so cited line numbers stay valid); the L–Z "15 files" list that has 16 entries and conflates three Status forms (re-verified per file: 10 no-header / 5 bold-wrapped / 1 bare); the 575-vs-592 bucket total; the W1b count; the "all five W1a shipped" overstatement; the retracted "no Pillar 6 in jc" claim; and 4609/11585 ≈ 0.40 (under half, not under a quarter). Gates: append_only_gate origin/main OK (9 files, none shrank); citation_decay --since origin/main 0 new / 147 pre-existing; SUPERSESSION-INDEX regenerated last, byte-identical. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp --- .claude/board/EPIPHANIES.md | 4 +- .claude/board/PLAN-INVENTORY-2026-09-07.md | 40 +++++++++++-------- .claude/board/TECH_DEBT.md | 2 +- .../plan-inventory-2026-09-07-plans-a-k.md | 7 +++- .../plan-inventory-2026-09-07-plans-l-z.md | 7 +++- ...inventory-2026-09-07-toplevel-and-board.md | 4 ++ ...lan-inventory-2026-09-07-w-waves-and-jc.md | 7 ++++ .claude/v3/COMPONENT-MAP.md | 2 +- .claude/v3/FUTURE-DESIGN.md | 2 +- .claude/v3/README.md | 11 +++-- .../v3/knowledge/sonnet-worker-guardrails.md | 4 +- 11 files changed, 61 insertions(+), 29 deletions(-) diff --git a/.claude/board/EPIPHANIES.md b/.claude/board/EPIPHANIES.md index 474c66de3..aa5d493b3 100644 --- a/.claude/board/EPIPHANIES.md +++ b/.claude/board/EPIPHANIES.md @@ -3,7 +3,7 @@ **Status:** FINDING, measured (five read-only Sonnet agents; every claim below re-verified by the orchestrator at a tree line, not a tag-file line). Full census: `.claude/board/PLAN-INVENTORY-2026-09-07.md`; evidence: `exec-runs/plan-inventory-2026-09-07-*.md`. **Confidence:** High for every `file:line` row; the D-id coverage percentage is measured on the A–K half of `.claude/plans/` only. -**The numbers.** 211 plans: 149 OPEN, 13 CLOSED, 9 SUPERSEDED, 40 AMBIGUOUS (35 with no status line at all — 22 of them the batch-landed `3DGS-*` design genre). V3 waves: W0 closed; W1 mechanism closed, adoption partial; W2 partial and self-contradictory; W3/W4 partial; W5/W6 open, W6's premise rescinded. W1a SIMD: 5/5 shipped. W1b: 0/6 closed (179 raw-intrinsic lines across five crates). W1.5: #6, #7 shipped and consumed, #8 unbuilt but ungated. +**The numbers.** 211 plans: 149 OPEN, 13 CLOSED, 9 SUPERSEDED, 40 AMBIGUOUS (35 with no status line at all — 22 of them the batch-landed `3DGS-*` design genre). V3 waves: W0 closed; W1 mechanism closed, adoption partial; W2 partial and self-contradictory; W3/W4 partial; W5/W6 open, W6's premise rescinded. W1a SIMD: 4/5 shipped, GATHER partial (API on every backend, x86 body a scalar polyfill). W1b: 0/5 TD entries closed, 1/7 files migrated (179 raw-intrinsic lines across five crates). W1.5: #6, #7 shipped and consumed, #8 unbuilt but ungated. **Direction one — docs still say Open for work on `main`.** Five `TECH_DEBT` rows said `Open` for primitives that have been on ndarray `master` for months (`simd_scalar.rs:1684,1709,1799,1887,1916`); W1.5 #7 said `Deferred` while `ndarray/src/hpc/randomized_signature.rs:292` ships it and `sigker/src/randomized.rs:124` consumes it — a THREE-way inconsistency, two knowledge docs having said SHIPPED since 2026-09-02; `STATUS_BOARD` D-LNC-5a/D-MW-P2 read `In PR #1198` with #1198 merged (`3797237b`); ENTROPY M2 said `QUEUED` nine weeks after D-PERT-1 shipped; `compiled-templates.md` said `StepMask` does not exist eight weeks after it did; `self-reasoning-substrate-v1.md:15` says "doc-only, no code" over four D-SRS rows that read Shipped with 147 tests between them. All but the last two are regraded in this PR (append-only strike-through with the tree line). @@ -60,7 +60,7 @@ ## 2026-09-07 — E-PLUG-AND-PLAY-IS-THE-DECLARATION-NOT-A-TABLE-1 — my fix rebuilt the lockstep it was closing -**Status:** FINDING, measured. Fixed in this PR (7 tests, both guards disable-verified). +**Status:** FINDING, measured. Fixed in the PR that landed this entry, #1216 (7 tests, both guards disable-verified). **Confidence:** High — the failure mode is demonstrated by a test that goes red when the scoping is restored. **Operator ruling (2026-09-07):** *"plug and play already has all the domains, diff --git a/.claude/board/PLAN-INVENTORY-2026-09-07.md b/.claude/board/PLAN-INVENTORY-2026-09-07.md index 7a9321ebc..6b40549d1 100644 --- a/.claude/board/PLAN-INVENTORY-2026-09-07.md +++ b/.claude/board/PLAN-INVENTORY-2026-09-07.md @@ -38,13 +38,13 @@ | `.claude/plans/*.md` L–Z (entries 104–211) | 108 | 85 | 12 | 7 | 4 | | **`.claude/plans/` total** | **211** | **149** | **13** | **9** | **40** | | `.claude/v3/` waves W0–W6 | 7 | 4 (W3/W4 partial, W5, W6) | 1 (W0) | — | 2 (W1 mechanism-closed/adoption-partial; W2 contradictory) | -| `.claude/v3/ENTROPY-MILESTONES.md` M1–M27 | 27 | 9 queued + 6 in-flight + 1 ruling-needed | ~8 | — | — | +| `.claude/v3/ENTROPY-MILESTONES.md` M1–M27 | 27 | 10 queued + 7 in-flight + 1 ruling-needed | 9 (M1 shipped with its gate tool queued; M2 regraded here) | — | — | | `.claude/*.md` top-level | 71 | — | — | 1 (`SESSION_FALKORDB_CROSSCHECK`) | 9 orphans, 61 pre-June-2026 | -| `STATUS_BOARD.md` D-id rows | 592 (89 sections) | 77 Queued, 23 Blocked, 5 In PR | 33 Shipped | 1 | 427 narrative-status cells | +| `STATUS_BOARD.md` D-id rows | 592 (89 sections) | 77 Queued, 23 Blocked, 5 In PR | 33 Shipped | 1 | 427 narrative-status cells (the nine leading-token buckets classify 575 rows; 17 rows carry a token outside them and were not classified — not a partition) | | `ISSUES.md` | 47 | 37 | 8 (+1 resolving) | 1 | — | | `TECH_DEBT.md` | 167 | 95 | 7 (4 Paid + 3 Resolved) | — | 65 unlabeled (pre-Kanban-format era) | -| W1a SIMD primitives (ndarray) | 5 | 0 | 5 | — | — | -| W1b consumer migrations | 6 | 6 (one half-done) | 0 | — | — | +| W1a SIMD primitives (ndarray) | 5 | 1 partial (#3 GATHER — API on all backends, x86 body a scalar polyfill) | 4 | — | — | +| W1b consumer migrations | 5 TD entries over 7 files | 5 TD entries (W2 half-done); 6 files | 0 TD entries; 1 file (`types.rs`, own path) | — | — | | W1.5 sigker primitives | 3 | 1 (#8, ungated) | 2 (#6, #7) | — | — | | jc pillar registry | 12 | 1 deferred (Pillar 2) + 1 feature-gated (Pillar 11) | 10 execute by default | — | — | @@ -66,13 +66,13 @@ recorded where it needs a reader. | Where | Said | Tree | Fixed here | |---|---|---|---| -| `TECH_DEBT.md` TD-NDARRAY-SIMD-{UNPACK-I4-16D, SATURATING-ABS-I8, GATHER, PREFETCH, POPCOUNT-U64} (W1a #1–#5) | `Status: Open` ×5 | every symbol on ndarray `master` `b9afcb9b` — `simd_scalar.rs:1684,1709,1799,1887,1916` / `simd_avx512.rs:2673,2716,2846,3091,2987` / `simd_neon.rs:2269,2301,2356,2445` | yes — regraded SHIPPED (primitive side); consumer halves stay Open | +| `TECH_DEBT.md` TD-NDARRAY-SIMD-{UNPACK-I4-16D, SATURATING-ABS-I8, GATHER, PREFETCH, POPCOUNT-U64} (W1a #1–#5) | `Status: Open` ×5 | every symbol on ndarray `master` `b9afcb9b` — `simd_scalar.rs:1684,1709,1799,1887,1916` / `simd_avx512.rs:2673,2716,2846,3091,2987` / `simd_neon.rs:2269,2301,2356,2445` | yes — four regraded SHIPPED (primitive side), GATHER regraded PARTIAL (API shipped, the AVX2 body the row requires is still open); consumer halves stay Open | | `TECH_DEBT.md` TD-NDARRAY-SIMD-RANDOMIZED-PROJECTION (W1.5 #7) | `DEFERRED` (heading + body) | `ndarray/src/hpc/randomized_signature.rs:96,205,244,292` (PR #294) **and consumed** — the import `use ndarray::hpc::randomized_signature::randomized_signature_sweep` at `crates/sigker/src/randomized.rs:49`; call site `RandomizedSignatureBuilder::encode` (`randomized.rs:124`) | yes — a three-way inconsistency (two knowledge docs said SHIPPED since 2026-09-02) | | `TECH_DEBT.md` TD-SIMD-SWEEP-W2 | "`types.rs` (22 raw ops)" | `types.rs` has **0** raw-intrinsic lines (`hamming_distance_dispatch` → `ndarray::hpc::bitwise::hamming_distance_raw`, `types.rs:462`); `ndarray_bridge.rs` still 66, incl. `_mm512_popcnt_epi64` at `:465,:493` | yes — HALF DONE | | `STATUS_BOARD.md` D-LNC-5a, D-MW-P2 | `**In PR** #1198` | #1198 merged `3797237b` | yes — Shipped | | `.claude/v3/ENTROPY-MILESTONES.md` M2 | `QUEUED (small, mechanical)` | STATUS_BOARD D-PERT-1 Shipped #630, verified in-code 2026-07-10 — nine weeks before the file was last edited | yes | | `.claude/v3/ENTROPY-MILESTONES.md` M24 | `PARTIAL 2026-07-17` | STATUS_BOARD D-MBX-A6-P3d/P4 (2026-08-01/02) ship the durable witness + recovery + cycle/WAL closure with the 64k/17 falsifier | yes — regrade appended | -| `.claude/v3/knowledge/compiled-templates.md` gap list | "`StepMask` type does not exist yet" | `contract::step_mask::StepMask`, D-V3-W3a, 2026-07-10 | yes | +| `.claude/v3/knowledge/compiled-templates.md` gap list; `COMPONENT-MAP.md` `StepMask` row ("BLOCKED/NEW — zero .rs matches"); `knowledge/sonnet-worker-guardrails.md` vocabulary row ("does not exist yet") | "`StepMask` type does not exist yet" | `pub struct StepMask` at `crates/lance-graph-contract/src/step_mask.rs:40`, exported `lib.rs:241`; D-V3-W3a, 2026-07-10 | yes — all three (the latter two after CodeRabbit on #1218) | | `.claude/plans/self-reasoning-substrate-v1.md:15` | "PROPOSED — doc-only. No code" | D-SRS-1..4 all **Shipped** on STATUS_BOARD (`reason.rs` + 7 tests; `shape,ancestry.rs` + 63; falsifier fired; `introspect.rs` + 77) | no — plan header, left for the owner; recorded | | `LATEST_STATE.md` static sections `## Current Contract Inventory (lance-graph-contract)` (`:2915`), `## Immediate Next Work` (`:3073`), `## Deferred (…)` (`:3102`) | frozen at 2026-05-16 / 2026-06-26; cite D-CSV-13b "IN PR" (shipped long since) and keep `TD-COLLAPSE-GATE-SMALLVEC-1` in "Deferred" while self-annotating it "CLOSED 2026-06-11" (`:3106`) | ~100 dated deltas prepended above them since | no — the file is append-only; a reader must treat the top-of-file deltas as current, the static sections as a 2026-06 snapshot | | `CLAUDE.md:762,765` | "`.claude/*.md` (61 top-level docs) … `SESSION_CAPSTONE.md`" | 71 files; no `SESSION_CAPSTONE.md` exists | no — CLAUDE.md is operator-owned; recorded | @@ -81,7 +81,7 @@ recorded where it needs a reader. | Where | Said | Truth | Fixed here | |---|---|---|---| -| `.claude/v3/INTEGRATION-PLAN.md:110` (D-CCF-4 row), `soa_layout/routing.md:104-106`, `knowledge/v3-substrate-primer.md:103-105` | `0x1000` is temporary; adoption 100 % ⇒ P4 ⇒ marker retires | **D-CCF-4 RESCINDED** (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) — STATUS_BOARD row 950 carried it for nine weeks; none of the three V3 files did | yes — ⊘ notes at all three | +| `.claude/v3/INTEGRATION-PLAN.md:110` (D-CCF-4 row), `soa_layout/routing.md:104-106`, `knowledge/v3-substrate-primer.md:103-105` | `0x1000` is temporary; adoption 100 % ⇒ P4 ⇒ marker retires | **D-CCF-4 RESCINDED** (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) — STATUS_BOARD row 950 carried it for nine weeks; none of the three V3 files did | yes — ⊘ notes at all three; the README five-sentence summary and the `sonnet-worker-guardrails.md` `0x1000` vocabulary row still said "temporary" and were corrected after CodeRabbit on #1218 | | `.claude/v3/INTEGRATION-PLAN.md` Addendum-12a (`:486-510`) vs Addendum-15 (`:744`) | W2a = a NEW gated `BoardAggregates` tenant **vs** W2a "SHIPPED as `ValueTenant::Kanban`" | `ValueTenant::Kanban` is the pre-existing per-ROW tenant #9 that `mailbox-kanban-model.md` calls a *sibling, not a substitute*; `tenants.md` (2026-08-23) lists no `BoardAggregates` | no — recorded on FUTURE-DESIGN 2026-09-07; needs `canonical_node.rs:1622+` + owner wiring read, not a doc pass | | `.claude/v3/COMPONENT-MAP.md` §4 W2b row | supervisor `KanbanActor` EXTEND | actor half DELETED 2026-08-05 (⊘ note already in the row since 2026-09-05); the wave table's W2b row still reads as an open work item | no — ⊘ exists; the wave table row is append-only history | | `persistent-nars-kg-v1.md`, `self-reasoning-substrate-v1.md` | cite `D-GRAPH-1`, `D-TRUTH-1`, `D-INFER-DEDUCTIONS-RELATION-BLIND` | **0** hits on STATUS_BOARD — never minted or renamed without a pointer | no — recorded | @@ -145,11 +145,13 @@ the two tag-files carry every row. | W5 consumer adoption | **OPEN** | nine sub-items, none Shipped; smb-office-rs `LanceConnector::upsert` still the orphan write (W5f) | | W6 monitor + retirement | **OPEN, premise RESCINDED** | D-CCF-4 rescinded 2026-07-03; W6a counting logic shipped, Lance-dataset sweep residue; "post-P4" now means post-checkpoint (⊘ notes landed) | -`ENTROPY-MILESTONES.md` (M1–M27 after this PR): shipped/resolved M2 (regraded +`ENTROPY-MILESTONES.md` (M1–M27 after this PR, all 27 accounted for): +shipped/resolved 9 — M1 (#628 fleet; its gate tool queued), M2 (regraded here), M5, M6, M7, M9 (partially reopened 2026-07-18), M15, M25, M26; in-flight -M3, M4, M12, M13, M17, M20, M24 (regraded here); queued M8, M10, M11, M14, M16, -M19, M21, M22, M23, M27 (mechanism shipped, callers queued); **M18 RULING-NEEDED** -(sigma chain Ω→Δ→Φ→Θ→Λ vs `KanbanColumn` six phases — no ruling found). +7 — M3, M4, M12, M13, M17, M20, M24 (regraded here); queued 10 — M8, M10, M11, +M14, M16, M19, M21, M22, M23, M27 (mechanism shipped, callers queued); **M18 +RULING-NEEDED** (sigma chain Ω→Δ→Φ→Θ→Λ vs `KanbanColumn` six phases — no +ruling found). 9 + 7 + 10 + 1 = 27. `knowledge/persona-vs-rung-ladder.md` O1–O9: **all open** except O5 (a probe run; its two hardenings unconfirmed). O2 re-scoped 2026-08-30 to "any-rung @@ -180,7 +182,9 @@ Landed in this PR: README doc-map row + collision note + two rulings added to the canonical list; primer §5 ⊘ (D-CCF-4) + §6 row (`E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1`, which postdated every mirror in the folder); INTEGRATION-PLAN W6 ⊘; routing.md §5 ⊘; -compiled-templates `StepMask` struck; ENTROPY M2/M24 regrades + M27; +compiled-templates `StepMask` struck (and, after CodeRabbit on #1218, the +COMPONENT-MAP `StepMask` row + the guardrails vocabulary row regraded +SHIPPED); ENTROPY M2/M24 regrades + M27; COMPONENT-MAP §6 `NestedBands` row; FUTURE-DESIGN 2026-09-07 block; witness-nibble-lane P5 (lit-harvest D3: "a single `u8:u8` rail read as ONE scalar axis is d=1", `:104`); the nexgen plan links back. **Deliberately not @@ -193,7 +197,7 @@ and a row without a lane is a wish (ENTROPY's own meta-rule). |---|---|---| | W1a #1 `from_i4_packed_u64` / `lane_i8` / `batch_packed_i4_16` | SHIPPED ×4 backends (scalar/avx512/neon + wasm partial) | TD said Open — fixed | | W1a #2 `saturating_abs` (I8x16/I8x32) | SHIPPED ×4 | TD said Open — fixed | -| W1a #3 `gather_u16` / `palette_lookup_u8x8` | SHIPPED — **API only**; x86 body is a scalar-loop polyfill by its own doc (`simd_avx512.rs:2833-2836`) | performance intent unmet | +| W1a #3 `gather_u16` / `palette_lookup_u8x8` | **PARTIAL** — API on scalar/avx512/neon; the x86 body is a scalar-loop polyfill by its own doc (`simd_avx512.rs:2833-2836`) | the AVX2 `_mm256_i32gather_epi32` body the TD row requires is still open; TD regraded PARTIAL, not SHIPPED | | W1a #4 `prefetch_read_t0/t1/t2` | SHIPPED ×3 | TD said Open — fixed | | W1a #5 `U64x8::popcnt` / `xor_popcount` / `U64x4::popcnt` | SHIPPED (scalar/avx512/avx2) | TD said Open — fixed | | W1b holograph `hamming.rs` | Open, 25 raw lines | — | @@ -208,7 +212,9 @@ and a row without a lane is a wish (ENTROPY's own meta-rule). Raw-intrinsic census across the five named crates: **179 lines** (the knowledge doc's "158-violation finding" of 2026-05-16 is stale as a live number — drift from growth, not remediation; `types.rs` is the one real -reduction). **W1b verdict: 0 of 6 closed** with all five primitives shipped. +reduction). **W1b verdict: 0 of 5 TD entries (W1–W5) closed — W2 half-done — +and 1 of 7 files (`types.rs`) migrated, 6 still raw**, with four primitives +shipped and GATHER partial. jc: the 12-entry registry is literally "11/12 implemented, Pillar 2 deferred" (`lib.rs:24-29`) — with the unstated caveat that Pillar 11 (`hambly_lyons`) @@ -243,8 +249,10 @@ run, not re-run here). `pattern.md` (17 refs) and `patterns.md` (30 refs) are two live entry points to one SoA/DTO doctrine. - **STATUS_BOARD.md**: 1854 lines, 89 sections, 592 rows in 3/4/6-column - shapes; 427 cells carry narrative rather than a lifecycle token — a per-row - mechanical census is not possible without normalizing the table shape. + shapes; 427 cells carry narrative rather than a lifecycle token, and the nine + leading-token buckets classify 575 rows — 17 rows carry a token outside them + and were not classified — so a per-row mechanical census is not possible + without normalizing the table shape. - **TECH_DEBT.md**: 167 entries under three header conventions; 65 unlabeled (all pre-Kanban-format, `TD-F10-ACTOR-ID`..`TD-INT-14`); open P0s after this PR: `TD-SIMD-SWEEP-W4`, `TD-API-DRIFT-MIDFLIGHT-1`, diff --git a/.claude/board/TECH_DEBT.md b/.claude/board/TECH_DEBT.md index 19d9a2998..b669c4d94 100644 --- a/.claude/board/TECH_DEBT.md +++ b/.claude/board/TECH_DEBT.md @@ -1787,7 +1787,7 @@ multi-antecedent path for real. - **Severity:** P1 (blocks `bgz17/src/simd.rs` migration off raw `_mm256_i32gather_epi32`) - **Surfaced in:** `simd-savant` PRE-MERGE audit 2026-05-16 (location: `crates/bgz17/src/simd.rs:88`) -- **Status:** ~~Open~~ **SHIPPED — API only, PARTIAL on intent.** Regraded 2026-09-07 (plan inventory): `U16x8::gather_u16` at `ndarray/src/simd_scalar.rs:1799` / `simd_avx512.rs:2846` / `simd_neon.rs:2356`, `palette_lookup_u8x8` at `simd_scalar.rs:1863` / `simd_avx512.rs:2900` / `simd_neon.rs:2423`. The x86_64 body is, by its own doc comment (`simd_avx512.rs:2833-2836`), "a scalar-loop polyfill (real AVX2 gather via `_mm256_i32gather_epi32` + downcast is tracked as a follow-up optimisation)" — so the contract (bounds-safe, three backends agree) is met and the performance intent of this row is not. Consumer half (`bgz17/src/simd.rs`, TD-SIMD-SWEEP-W3) still Open and still calls both gathers raw. +- **Status:** ~~Open~~ **PARTIAL — API shipped on all three backends; the AVX2 gather body this row requires is still open.** Regraded 2026-09-07 (plan inventory; leading token corrected from SHIPPED to PARTIAL after CodeRabbit on #1218): `U16x8::gather_u16` at `ndarray/src/simd_scalar.rs:1799` / `simd_avx512.rs:2846` / `simd_neon.rs:2356`, `palette_lookup_u8x8` at `simd_scalar.rs:1863` / `simd_avx512.rs:2900` / `simd_neon.rs:2423`. The x86_64 body is, by its own doc comment (`simd_avx512.rs:2833-2836`), "a scalar-loop polyfill (real AVX2 gather via `_mm256_i32gather_epi32` + downcast is tracked as a follow-up optimisation)" — so the contract (bounds-safe, three backends agree) is met and the performance intent of this row is not. Consumer half (`bgz17/src/simd.rs`, TD-SIMD-SWEEP-W3) still Open and still calls both gathers raw. - **Description:** `bgz17` uses `_mm256_i32gather_epi32` directly for palette lookup (8 u16 indices → values). `ndarray::simd` exposes no gather primitive — the polyfill needs `U16x8::gather_u16` or a dedicated `palette_lookup_u8x8` helper. Dominant palette-stream workload primitive; missing it forces every palette consumer to reinvent gather. - **Required API surface:** - `impl U16x8 { pub fn gather_u16(indices: U16x8, table: &[u16]) -> Self; }` — AVX2 `_mm256_i32gather_epi32` + downcast; NEON scalar loop (no native gather); scalar `indices.iter().map(|&i| table[i])`. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md index 9258ea7a4..bb6516931 100644 --- a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-a-k.md @@ -333,4 +333,9 @@ mechanical pass, or remain genuinely uncertain after it: No file was left completely unclassified — every row in the table above has a verdict. The AMBIGUOUS verdicts (36 total) are 35 "no status line found" cases plus the 1 `cascade-seal-register-grid-v1.md` case above; none are -unexplained gaps. \ No newline at end of file +unexplained gaps. + +## Orchestrator errata (appended 2026-09-07 after review of #1218 — the agent text above is unedited) + +- L44 (`belief-abi-step2-addendum-copula-v1.md`): the `status_token` cell stops at an opening backtick. The plan's status (its lines 3–5) reads in full: "ADDENDUM to" the path `.claude/plans/belief-abi-step2-ruling-request-v1.md`, "operator-directed 2026-08-23. **Proposes NO mint.** Supersedes the ruling request's Item 1 recommendation, which is retracted below." Verdict OPEN unchanged. +- L119, L196, L294: `## Mismatches`, `## Candidate epiphanies`, `## Unread / uncertain` were written without a preceding newline (a write artifact). The text is intact; the file is deliberately not re-flowed so the line numbers cited in the review threads and here stay valid. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md index 2a6b60c98..9505994b9 100644 --- a/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-plans-l-z.md @@ -577,4 +577,9 @@ Facts observed this pass, with counts and filenames, no speculation: file's §0-§4 and §7 (self-assessment) were read; §5 ("Proof of Vision") and §6 ("Cross-References") were not read in full. The OPEN verdict rests on §7's own admission of open items, which should hold regardless, but a - full read of §5/§6 was not performed. \ No newline at end of file + full read of §5/§6 was not performed. + +## Orchestrator errata (appended 2026-09-07 after review of #1218 — the agent text above is unedited) + +- L39 (`lance-convergence-staged-migration-v1.md`): the `status_token` cell copies only the first line of a wrapped status. In full (plan lines 3–4): "MEASURED / ready-to-execute for stages 0–1; stage 3 is PROBE-GATED." Verdict OPEN unchanged. +- L518–533 (finding 7): the list has 16 entries, not 15, and the heading conflates three forms. Re-verified 2026-09-07 against the files: (a) 10 with no `Status:` header of any form — ogit-g-context-bundle-v1, probe-revision-attention-view-1, unified-soa-rubikon-integration-v1 (a "Status legend" only), v3-substrate-integration-v1, measure-64k-axes-v1/v2/v3/v4, soa-migration-diff-resolution-2026-06-13 (only a section-level "Status in lance-graph"), unified-ogit-architecture-v1 (per-section "Status." lines only); (b) 5 whose header IS bold but wraps the value — `**Status: VALUE**` or `> **Status:** VALUE` — so a `\*\*Status:?\*\*` regex misses it: token-value-tenant-v1, mul-ewa-trust-propagation-v1, r2il-bpe-typed-genetic-recombination-v1, soa-value-tenant-migration-v1-harvest, wikidata-lazy-spine-hydration-v1 — the "bare `Status:`, no `**`" description of these five above is wrong; the regex-miss is real, the bareness is not; (c) 1 genuinely bare label: tarski-markov-hhtl-seam-v1 (`> Status: **HELD …`). The L–Z verdicts are unaffected; the four L–Z AMBIGUOUS files in the inventory §0 are ogit-g-context-bundle-v1, probe-revision-attention-view-1, supabase-subscriber-v1, tarski-markov-hhtl-seam-v1. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md index 6e09885e3..1e3871c77 100644 --- a/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-toplevel-and-board.md @@ -646,3 +646,7 @@ file:line anchor, offered as raw material — not claimed as ratified epiphanies had no single unambiguous "OPEN"/"RESOLVED" sentence — the OPEN classification there is inferred from prose ("Not root-caused or fixed... Not scheduled") rather than a literal status field. + +## Orchestrator errata (appended 2026-09-07 after review of #1218 — the agent text above is unedited) + +- L150–160: the nine leading-token buckets sum to 575 of the 592 data rows; the remaining 17 rows carry a leading token outside those nine classes and were not classified in this pass. Read the table as "575 classified + 17 unclassified = 592", not as a partition. Mirrored in the inventory §0 / §6. diff --git a/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md b/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md index 2679d0ae5..eed844a84 100644 --- a/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md +++ b/.claude/board/exec-runs/plan-inventory-2026-09-07-w-waves-and-jc.md @@ -264,3 +264,10 @@ orchestrator to promote into a board entry if it chooses: - **The four un-registered jc example files** (`l9_loci_real_text.rs`, `partof_isa_vs_palette256.rs`, `rung_divergence_reliability.rs`, `weather_substrate_reliability.rs`) were confirmed present by `ls` and confirmed absent from `Cargo.toml`'s `[[example]]` list by full read, but were not opened themselves — unknown whether they compile cleanly under default features, and unknown whether Cargo's implicit auto-discovery (rather than an explicit entry) is intentional or an oversight. - **`neural-debug`, `lance-graph-planner`, and other workspace crates were not grepped for raw intrinsics** — the task scoped the raw-intrinsic census to the 5 crates named in the knowledge doc's per-workload table (holograph, lance-graph/blasgraph, bgz17, lance-graph-contract, thinking-engine); a wider sweep might turn up additional violations outside that named set. - **The `is_x86_feature_detected!`/`is_aarch64_feature_detected!` grep pattern was included in the raw-intrinsic count** per the task's own symbol list; this inflates the "violation" count slightly relative to a strict "raw SIMD intrinsic call" definition (e.g., `bgz17/src/simd.rs`'s `detect_simd()` calls are architecturally a duplicate-dispatch anti-pattern, AP-SIMD-8, rather than a raw-instruction call per se — both are real violations per the workspace's own `simd-savant` doctrine, just different flavors). + +## Orchestrator errata (appended 2026-09-07 after review of #1218 — the agent text above is unedited) + +- L61–64 (W1b verdict): "0 of 6 consumer-migration TD entries" mixes two counts. There are 5 TD entries (TD-SIMD-SWEEP-W1..W5) and 7 files in the table. Correct reading: 0 of 5 TD entries closed (W2 half-done); 1 of 7 files migrated (`types.rs`); 6 files still carry raw intrinsics. +- L81–84 item 1 and L248–250 item 1: "all five W1a entries are shipped work with stale labels" overstates #3. Item 3 of the same list already has it right — `gather_u16` ships as a scalar-loop polyfill on x86 — so TD-NDARRAY-SIMD-GATHER is PARTIAL (API on all backends; the AVX2 gather the entry requires is still open), not SHIPPED. `TECH_DEBT.md` and the inventory carry the PARTIAL grade. +- L253 item 6: retracted — jc DOES have a Pillar 6: `crates/jc/src/ewa_sandwich.rs` labels itself so and `lib.rs` numbers the same module 9. An internal jc numbering conflict, not another document's numbering (Codex on #1218; inventory §5/§7). +- L255 item 8: "well under a quarter" is arithmetic drift — 4609 / 11 585 ≈ 0.40, under half, not under a quarter. The conclusion (W5 legitimately deferred by its own trigger) stands. diff --git a/.claude/v3/COMPONENT-MAP.md b/.claude/v3/COMPONENT-MAP.md index 079489331..b25cc61a7 100644 --- a/.claude/v3/COMPONENT-MAP.md +++ b/.claude/v3/COMPONENT-MAP.md @@ -91,7 +91,7 @@ Meta/Plasticity widths differ persisted-vs-hot with no parity test; | template-equivalence (Exact/RankOrder live; Semantic fail-closed) | REUSE | the W3 merge gate; honest deferred grading | | cognitive-compiler ScaffoldCompiler | REUSE | "no trace → no template" enforced in types; synthesis = first probe | | `FieldMask` (class_view.rs:69) | REUSE | the sibling StepMask mints next to (u64 position bitmask) | -| `StepMask` | BLOCKED/NEW | zero .rs matches — docs-only; D-V3-W3a | +| `StepMask` | ~~BLOCKED/NEW~~ **SHIPPED** (regraded 2026-09-07, plan inventory) | ~~zero .rs matches — docs-only~~ `pub struct StepMask` at `crates/lance-graph-contract/src/step_mask.rs:40`, exported `lib.rs:241`; D-V3-W3a shipped 2026-07-10; W3b/c/d (the adapter that consumes it) still queued | | graph-flow NextAction | REUSE | **6 variants incl. no-op GoBack** (docs said 5 — corrected) | | **NextAction ↔ OgarAction "1:1"** | CORRECTED | the honest 1:1 is `Step ↔ Task` and `ogar_name() ↔ Task::id()`; Continue/WaitForInput/GoTo/End have NO template-side counterparts — closing this IS the StepMask/adapter work, see compiled-templates.md correction | | rs-graph-llm `template-task` crate | EXTEND | the adapter's natural home — Task shims exist with literal placeholders; ZERO Cargo dep on lance-graph yet (episodic-arc-task proves the git-dep pattern) | diff --git a/.claude/v3/FUTURE-DESIGN.md b/.claude/v3/FUTURE-DESIGN.md index 00f6a556d..000f0bc77 100644 --- a/.claude/v3/FUTURE-DESIGN.md +++ b/.claude/v3/FUTURE-DESIGN.md @@ -19,7 +19,7 @@ | `E-EVERYTHING-WIRES-TO-SOA-V3-CE64-IS-ALU-LEGACY-1` (operator 2026-09-05) | "anything must be wired into SoA V3 substrate no exceptions except the causaledge64 adjacent as ALU legacy substrat" — landed the day COMPONENT-MAP was last edited and was cited nowhere in this folder; now primer §6 last row + README ruling list | every BLOCKED/NEW row in COMPONENT-MAP is read through it | | `D-CCF-4` RESCINDED (operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`) | STATUS_BOARD row carried the rescission for nine weeks while INTEGRATION-PLAN W6, `routing.md` §5 and primer §5 still described `0x1000` as temporary with a 100 %-adoption retirement trigger; ⊘ notes now at all three | W6b/W6c "post-P4" = post operator-checkpoint on the corpus proof, never post-retirement | | W2a is stated two ways in one file — RECORDED, not adjudicated | INTEGRATION-PLAN Addendum-12a (2026-07-02): a NEW gated `BoardAggregates` tenant behind T1–T6 + a batched mint. Addendum-15 (2026-08-04): "SHIPPED as `ValueTenant::Kanban`" — the pre-existing per-ROW tenant #9 that `mailbox-kanban-model.md` calls a sibling, never a substitute; `tenants.md` (refreshed 2026-08-23) lists no `BoardAggregates`. Which is current needs `canonical_node.rs:1622+` and the owner wiring read, not a doc-only pass | owner-session / operator call; until then W2a reads PARTIAL, and M24's residue stays gated on it | -| `compiled-templates.md` gap list said `StepMask` "does not exist yet" | It shipped 2026-07-10 (`contract::step_mask::StepMask`, D-V3-W3a); the doc lagged eight weeks. Struck 2026-09-07 | W3b/c/d still Queued | +| `compiled-templates.md` gap list said `StepMask` "does not exist yet" — and so did the `COMPONENT-MAP.md` `StepMask` row ("BLOCKED/NEW — zero .rs matches") and the `sonnet-worker-guardrails.md` vocabulary row | It shipped 2026-07-10 (`contract::step_mask::StepMask`, D-V3-W3a); the docs lagged eight weeks. Struck / regraded 2026-09-07 at all three (the latter two after CodeRabbit on #1218) | W3b/c/d still Queued | ## 2026-07-17 (sixth wave) — ACK ELIMINATED everywhere (operator directive) diff --git a/.claude/v3/README.md b/.claude/v3/README.md index 02e696061..1942b44c2 100644 --- a/.claude/v3/README.md +++ b/.claude/v3/README.md @@ -21,9 +21,12 @@ orchestration what askama × FieldMask is to rendering; a successful LLM (Rig oracle) run compiles DOWN to a deterministic replayable template executed by graph-flow under inherited ownership. Thinking cycles follow **standing async plans** within a 550 ms net budget; updates reprioritize, -never gate. The `0x1000` custom half is a temporary **adoption monitor**; -at 100% adoption (P4) it retires and the classview half opens for the 64k -render/template catalogue. +never gate. The `0x1000` custom half is a **permanent adoption monitor** — +~~temporary; at 100% adoption (P4) it retires~~ (`D-CCF-4` RESCINDED, +operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`; corrected here +2026-09-07): P4 at 100% adoption is an operator **checkpoint** on the corpus +proof, never a retirement, and the classview half's 64k render/template +catalogue (W6b/c) is gated on that checkpoint. ## Doc map (read order per task) @@ -43,7 +46,7 @@ render/template catalogue. | waking an agent | `agents/BOOT.md` (the four `v3-*` cards + trigger routing) | | landing future design / post-V3 rulings | `FUTURE-DESIGN.md` (the meta board: ruling index, migration arc, thinking-engine gem wiring queue) | | keeping the vision (the WHY, graded) | `VISION.md` — the AGI-aspiring canon re-grounded 2026-07-10; every claim [G]/[RULING]/[ASPIRATION]; filigree-reviewed | -| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12; the plan's sealed object is keyed `(classid, version)` — THIS folder's keyspace — as its PROPOSED room-26 shape, while the shipped D-NXG-1 `NestedBands` is version-keyed over one column with no classid; its room 27 proposes it as the carrier primer §6 row 3 demotes `Vsa16kF32` from) + board `E-NXG-*` (2026-09-05) + `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02: of the loco calls it treated as constitutional only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory) — before that neither doc family cited the other | +| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12). Two facts to hold apart: (a) the plan's PROPOSED room-26 shape keys its sealed object `(classid, version)` — THIS folder's keyspace — while the shipped D-NXG-1 `NestedBands` is version-keyed over one column with no classid; (b) its room 27 proposes `NestedBands` as the carrier that primer §6 row 3 demotes `Vsa16kF32` from. Companions: board `E-NXG-*` (2026-09-05) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02 — of the loco calls it treated as constitutional, only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory); before that neither doc family cited the other | Shortcuts: `/v3` (bootload), `/v3-audit` (pre-commit conformance greps). Canonical ruling texts live on the board (`.claude/board/EPIPHANIES.md`, diff --git a/.claude/v3/knowledge/sonnet-worker-guardrails.md b/.claude/v3/knowledge/sonnet-worker-guardrails.md index 10e1780e1..0e6e59b00 100644 --- a/.claude/v3/knowledge/sonnet-worker-guardrails.md +++ b/.claude/v3/knowledge/sonnet-worker-guardrails.md @@ -73,10 +73,10 @@ exactly the row below, never a neighbor. | **BindSpace** | the legacy global store, MIGRATION IN PROGRESS → MailboxSoA | something to extend or to delete now | add new writers to it; remove it | | **baton / emission / CollapseGateEmission / emit()** | tombstoned concepts (removed from source) | anything to restore | reintroduce, even "for compatibility" | | **owner / on_behalf** | `SoaEnvelope::mailbox_owner()` stamp + batch-writer cast pairing | a DTO field | add ownership fields to DTOs (§1 rule 9) | -| **0x1000 (custom half)** | the V3-adoption MONITOR marker (temporary by declaration) | a domain id or a semantic flag | branch business logic on it; mint new meanings for it | +| **0x1000 (custom half)** | the V3-adoption MONITOR marker — permanent (`D-CCF-4` RESCINDED, operator 2026-07-03, `E-V3-DUAL-SCHEMA-0x1000-IS-PERMANENT-1`; ~~temporary by declaration~~ regraded 2026-09-07) | a domain id or a semantic flag; something that retires at 100 % adoption | branch business logic on it; mint new meanings for it; plan its retirement | | **canon / custom** | classid halves: hi u16 = canon concept, lo u16 = custom marker/render | "old/new" | read halves with bit math (§1 rule 4) | | **template** | a compiled, replayable orchestration artifact (elixir-template DSL) | a prompt or a string | degrade a template into a prompt | -| **StepMask** | a QUEUED contract type (W3a) — does not exist yet | something to invent ad hoc | mint it without the W3a spec | +| **StepMask** | `contract::step_mask::StepMask` — SHIPPED 2026-07-10 (D-V3-W3a); ~~a QUEUED contract type (W3a) — does not exist yet~~ (regraded 2026-09-07, plan inventory) | something to invent ad hoc, or to re-mint | re-mint it; W3b/c/d consume the shipped type | | **facet** | the V3 16-byte atom: 4B prefix (domain\|appid\|classview) + 96-bit payload (soa_layout/le-contract.md §3 catalogue L1–L8) | a free-form struct | add a layout outside the L1–L8 catalogue; put a label/position in a slot | | **classview (lo u16)** | the ClassView selector in the classid custom half — labels + positions resolve THROUGH it | a place to store data bits | branch on raw classview values other than via read-mode helpers | | **rail** | a 6×(8:8) payload plane of one-byte refs (part_of:is_a etc.) | an edge list to grow | change rail arity/stride | From 14047c357f97fa0646f9b0a4c09d28f25e987a05 Mon Sep 17 00:00:00 2001 From: Claude Date: Mon, 7 Sep 2026 16:22:24 +0000 Subject: [PATCH 4/4] =?UTF-8?q?board:=20CodeRabbit=20round-2=20on=20#1218?= =?UTF-8?q?=20=E2=80=94=20`#1218`=20no=20longer=20at=20a=20line=20start=20?= =?UTF-8?q?(MD018);=20README=20room-27=20clause=20restated=20without=20the?= =?UTF-8?q?=20stranded=20preposition?= MIME-Version: 1.0 Content-Type: text/plain; charset=UTF-8 Content-Transfer-Encoding: 8bit Two follow-ups from CodeRabbit's second review (of 162963a5): - PLAN-INVENTORY §5: the sentence "…Codex review on / #1218 caught it." wrapped so that `#1218` opened a line, which markdownlint reads as a malformed heading (MD018). Reflowed; text unchanged. - .claude/v3/README.md doc-map row: "the carrier that primer §6 row 3 demotes `Vsa16kF32` from" was read as truncated. Restated as "the `Think.trajectory` carrier in place of `Vsa16kF32`, which primer §6 row 3 demotes out of that carrier role" — same claim, no stranded preposition. - The other round-2 finding (EPIPHANIES line 47 "altered board history") is declined: the paragraph belongs to an entry THIS PR adds, not a historical entry on main; in-place edits to content created within the same unmerged PR are permitted. Gates: append_only_gate origin/main OK; citation_decay --since origin/main 0 new; SUPERSESSION-INDEX regenerated last, byte-identical. Co-Authored-By: Claude Fable 5.1 Claude-Session: https://claude.ai/code/session_01KCGhDYoQBXs3poaR7sFuqp --- .claude/board/PLAN-INVENTORY-2026-09-07.md | 4 ++-- .claude/v3/README.md | 2 +- 2 files changed, 3 insertions(+), 3 deletions(-) diff --git a/.claude/board/PLAN-INVENTORY-2026-09-07.md b/.claude/board/PLAN-INVENTORY-2026-09-07.md index 6b40549d1..51de4540e 100644 --- a/.claude/board/PLAN-INVENTORY-2026-09-07.md +++ b/.claude/board/PLAN-INVENTORY-2026-09-07.md @@ -228,8 +228,8 @@ modules 9 / 9b (`9. EWA-sandwich Σ-push-forward` at `crates/jc/src/lib.rs:16`). The board's "Pillar-6/7" rows (D-NXG-11, D-MEP-0) and `mul-ewa-trust-propagation-v1` follow the module-doc numbering — they DO mean jc's EWA sandwich, and the collision is internal to jc. (The W-waves tag-file -and this document's first draft said "no Pillar 6 in jc"; Codex review on -#1218 caught it.) `jc::solver_order` is a 13th battery outside +and this document's first draft said "no Pillar 6 in jc"; Codex review on #1218 +caught it.) `jc::solver_order` is a 13th battery outside `run_all_pillars()`. Four example files are present with no `[[example]]` entry (`l9_loci_real_text`, `partof_isa_vs_palette256`, `rung_divergence_reliability`, `weather_substrate_reliability`). W5's diff --git a/.claude/v3/README.md b/.claude/v3/README.md index 1942b44c2..4e9e9d03f 100644 --- a/.claude/v3/README.md +++ b/.claude/v3/README.md @@ -46,7 +46,7 @@ catalogue (W6b/c) is gated on that checkpoint. | waking an agent | `agents/BOOT.md` (the four `v3-*` cards + trigger routing) | | landing future design / post-V3 rulings | `FUTURE-DESIGN.md` (the meta board: ruling index, migration arc, thinking-engine gem wiring queue) | | keeping the vision (the WHY, graded) | `VISION.md` — the AGI-aspiring canon re-grounded 2026-07-10; every claim [G]/[RULING]/[ASPIRATION]; filigree-reviewed | -| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12). Two facts to hold apart: (a) the plan's PROPOSED room-26 shape keys its sealed object `(classid, version)` — THIS folder's keyspace — while the shipped D-NXG-1 `NestedBands` is version-keyed over one column with no classid; (b) its room 27 proposes `NestedBands` as the carrier that primer §6 row 3 demotes `Vsa16kF32` from. Companions: board `E-NXG-*` (2026-09-05) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02 — of the loco calls it treated as constitutional, only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory); before that neither doc family cited the other | +| row masks / histograms / Prozentrang / rank over classid-addressed, version-keyed slabs | `.claude/nexgen/harvest/00-INDEX.md` → `.claude/nexgen/plans/nexgen-mask-histogram-thresholds-v1.md` (D-NXG-1..12). Two facts to hold apart: (a) the plan's PROPOSED room-26 shape keys its sealed object `(classid, version)` — THIS folder's keyspace — while the shipped D-NXG-1 `NestedBands` is version-keyed over one column with no classid; (b) its room 27 proposes `NestedBands` as the `Think.trajectory` carrier in place of `Vsa16kF32`, which primer §6 row 3 demotes out of that carrier role. Companions: board `E-NXG-*` (2026-09-05) and `.claude/knowledge/literature-harvest-2026-09-01-post-1132.md` (partially superseded 2026-09-02 — of the loco calls it treated as constitutional, only `TERNLOG 0x86` survives). Cross-linked 2026-09-07 (plan inventory); before that neither doc family cited the other | Shortcuts: `/v3` (bootload), `/v3-audit` (pre-commit conformance greps). Canonical ruling texts live on the board (`.claude/board/EPIPHANIES.md`,