Cycle Compression Foundry

Read the 13-domain sweep · Submit a buyer-controlled replay candidate · Platform pricing and checkout

The pricing link sells US Tech Automations' general workflow platform. No cycle-compression SKU is commercially admitted or available for checkout while the evidence decision remains RETURN_NONE.

Demand first: prospective buyers can review the inbound qualification contract and, on the public repository, submit the structured Cycle compression demand issue. Never attach obligations, ledger exports, contracts, or personal data.

Project 67 is an evidence-gated foundry for discovering one narrow, repeatable class of pre-settlement obligation cycles under one controller. A possible future product would propose fewer payment and journal instructions while preserving every party's final position.

This repository is an offline research and control system. It is not a payment processor, clearinghouse, accounting system, recovery tool, reconciliation product, legal adviser, counterparty, creditor, or custodian.

Implemented scope

• Phase 0 charter, authority matrix, exclusions, economics, evidence policy, prompt contracts, and autonomy boundaries;
• typed source-receipt and domain-card contracts;
• a deterministic, fail-closed Phase 1 admission engine;
• a public-source sweep of thirteen unrelated domains;
• an immutable obligation-snapshot contract and strict eligibility filter;
• deterministic directed-cycle cancellation with exact compatibility partitions;
• a separately implemented invariant verifier and gross-settlement replay;
• decimal-only economics and the 70% / 10× / <$2k gates;
• a ten-state idempotent workflow on a SHA-256-chained SQLite event store;
• backup/restore, fault injection, prompt pinning, safe-pause, and readiness controls;
• a zero-spend model harness and settled-cash token budget;
• a locked finite autopilot that exits safely when no eligible work exists;
• a 100-run synthetic end-to-end transaction twin; and
• a governed public prospect-to-diagnostic-draft lane with explicit cold-versus-warm status, exact published contact proof, and an idempotent review-only Google Sheets projection;
• a private, receipt-gated bridge that prepares sanitized directions and respects USTA's upstream trusted-commercial-proof admission gate; and
• an isolated Hermès no-agent canary with public quote-anchor reproduction, one-use exact authorization, stable Sheets replay receipts, USTA immutability proof, strict value-contract telemetry, and automatic job removal; and
• tests for evidence, policy, graph, replay, state, storage, budget, recovery, outreach gates, Sheets conflicts, and global RETURN_NONE behavior.

Factual status

• Phase 0 controls: implemented; operator ratification and specialist panel approvals are pending.
• Phase 1 public prescreen: implemented. Its current decision is RETURN_NONE.
• Candidate-independent shadow transaction twin: qualified. Eleven synthetic gross obligations compress to two proposed payments, independent replay preserves every node, and 100/100 unattended reruns match.
• Odoo multi-company accounting remains a research lead, not an admitted domain. Public evidence does not prove two settled paying customers for the exact product, buyer-specific setoff authority, a quantified exact-class replay, or the operating targets.
• Private prospect packets, contact evidence, drafted outreach, review-sheet state, and production-integration receipts are intentionally excluded from the public release. No outbound message is authorized by this repository.
• Buyer historical replay, marketplace listing, message transmission, live installation, revenue, and self-funding are closed. Candidate-independent controls for those later states are prepared and tested with synthetic receipts only.

RETURN_NONE is a successful safety result, not a claim that no opportunity can ever exist. The sweep can be rerun when materially stronger receipts exist.

Commands

PYTHONPATH=src python3 -m cycle_compression_foundry doctor
PYTHONPATH=src python3 -m cycle_compression_foundry validate \
  research/phase1/sweep-001-public-prescreen.json
PYTHONPATH=src python3 -m cycle_compression_foundry decide \
  research/phase1/sweep-001-public-prescreen.json \
  --output state/latest-phase1-decision.json
PYTHONPATH=src python3 -m cycle_compression_foundry snapshot-validate \
  fixtures/synthetic/cycle-snapshot-001.json
PYTHONPATH=src python3 -m cycle_compression_foundry qualify --state-dir state
PYTHONPATH=src python3 -m cycle_compression_foundry readiness --state-dir state
PYTHONPATH=src python3 -m cycle_compression_foundry autopilot --state-dir state
PYTHONPATH=src python3 -m unittest discover -s tests -v

Passing tests proves the controls behave as implemented. It does not prove legal authority, graph feasibility in a buyer's ledger, demand, savings, unattended operation, production readiness, or revenue.