Repository navigation
fix(forge-friction): classify themes as operational vs shape-gap before routing to forge - #13
Merged
Merged
Conversation
…re routing to forge The friction analyzer surfaces tool errors from session telemetry, but almost all of them are operational noise (timeouts, rate limits, usage caps, hook blocks) rather than genuine workflow shape gaps that would make useful skills. This caused /forge-friction to propose skills for infrastructure issues that should be hooks or config changes. Add a 3-step pipeline: Filter, Classify, Route. - Step 1 (Filter): unchanged confidence-class filtering - Step 2 (Classify): each theme is tagged operational or shape-gap using explicit indicator lists (timeouts, rate limits, hook blocks, connection errors, model rejections, iteration caps = operational; recurring ad-hoc multi-step workarounds without skill coverage = shape-gap) - Step 3 (Route): operational themes get a one-line recommendation (hook/config/budget fix); only shape-gap themes proceed to /forge Dry-run mode now outputs two separate tables. Auto mode skips operational themes for forge. Interactive mode lists operational recommendations first (informational, no y/n), then shape-gap candidates (y/n to forge).
…ble language from Auto mode - F2: $ARGUMENT → $ARGUMENTS on line 21 (singular form silently disabled --dry-run gate) - F3: Auto mode intro scoped to Step 2 shape-gap classification pipeline, removing stale skill-addressable filter criteria
This was referenced Oct 24, 2026
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Problem
/forge-frictiontreats every telemetry theme as a skill candidate. In practice, almost all themes are operational noise: bash timeouts (235 sessions), rate limits (37 sessions), usage caps (90 sessions), hook blocks (523 sessions). None of these map to useful skills. Running/forge-frictiontoday produces zero actionable skill candidates.Fix
Add a 3-step pipeline to the SKILL.md: Filter -> Classify -> Route.
Step 2 (Classify) is new. Each theme is tagged as one of:
Step 3 (Route) sends each type to the right output:
/forgedispatch (interactive y/n or auto)Changes
skills/forge-friction/SKILL.md: restructured "Your job" into Steps 1-3, added classification criteria, updated dry-run to show two tables, updated auto/interactive flows to only forge shape-gaps, updated summary formatEvidence
Ran
/forge-frictionon 1,569 friction sessions. The analyzer found categories likebash:timeout(235 sessions),hook_block:bash(523 sessions),subagent_fail:UsageLimitError(90 sessions). All are operational. Zero shape-gap themes surfaced from telemetry alone, which is correct: the telemetry source inherently produces operational signal. The classifier prevents these from reaching/forge.