fixlisted

# Fix ## Success I := ∀ f ∈ actionable → applied ∨ deferred (issue ∃) ∨ skipped (user) ∧ PR comment posted V := `gh pr view {N} --comments | grep "## Review Fixes Applied"` Two-pass pipeline: auto-apply high-C findings (C≥T, 2+ agents), then 1b1 for rest. **⚠ Continuous pipeline. ¬stop between phases. Stop only on: unrecoverable failure or Phase 8 completion.** ``` /fix → findings from conversation context /fix #42 → gather findings from PR #42 comments ``` ## Pipeline | Phase | ID | Required | Verifies via | Notes | |-------|----|----------|---------------|-------| | 1 | gather | ✓ | F parsed | — | | 2 | triage | ✓ | Q_auto + Q_1b1 split | — | | 3 | auto-apply | — | applied count | Q_auto = ∅ → skip | | 4 | push-auto | — | `git push` success | ¬applied → skip | | 5 | walkthrough | — | decisions recorded | Q_1b1 = ∅ → skip | | 6 | apply-1b1 | — | applied count | acc = ∅ → skip | | 7 | final-push | ✓ | `git push` success | — | | 8 | post-comment | — | comment posted | ∄ PR → skip | ## Pre-flight Success: ∀ actionable → applied ∨ deferred ∧ PR comment posted Evidence: `gh pr view {N} --comments | grep "## Review Fixes Applied"` Steps: gather → triage → auto-apply → walkthrough → apply-1b1 → final-push → post-comment ¬clear → STOP + ask: "Do you have review findings to fix?" Let: F := all findings | f ∈ F | C(f) ∈ [0,100] ∩ ℤ — confidence A(f) := {agents that flagged f} | cat(f) ∈ {issue, suggestion, todo, nitpick, thought, question, praise} src(f) := o