daily-fix: crash-fix relaunch asserts fix commit on pod + wi
kind: infra#wf-fix#wf-fix-fp:22f16a77ea22#daily-auto-filed
track:Highlight text on any card (body, plan, or an event) to anchor a comment, or leave a whole-task comment here. Mention @claude to summon a reply.
No comments yet.
Overview / Motivation
Auto-filed by /daily 2026-07-05 (route 2: behavior/logic change -> independent review), from the nightly transcript problem sweep.
Goal
Extend the crash-fix-rounds fix-engaged-signal contract: after a code-fix round, the relaunch (a) asserts the specific fix commit is present on the pod (git HEAD / merge-base probe) before dispatch, and (b) wipes checkpoints written by the stale-code run so a resume cannot inherit poisoned state; name both in the declared fix-engaged signal.
Workflow gap
- Bug observed: #779 stale-code launch: launch #1 ran the commit BEFORE the implementer's val-lambda fix, hit the known GCV ridge degeneracy at n~d (val R^2 -4.7 vs ~0.6 expected) and checkpointed the garbage; the restart faithfully resumed the poisoned checkpoints. Recovery required wiping checkpoints and relaunching at current HEAD with a 75s health probe reporting pod git HEAD + first val-selection lines. Session 5664c4f8, task #779, 2026-07-06.
- Why it is a workflow gap: the failure originates in the workflow surface / shared helper named below, not in any one experiment.
- Confidence (emitter): medium
Scope / surfaces
- Primary target:
.claude/rules/crash-fix-rounds.md - The experimenter pre-launch sync mandate exists; this adds the assert-the-fix-commit + checkpoint-hygiene specifics for crash-fix rounds.
Constraints / invariants
scripts/workflow_lint.py --check-asks+--check-referencesstay green; ruff clean on touched files; relevant tests pass.
Provenance
- workflow_fix_target: .claude/rules/crash-fix-rounds.md
- source: /daily 2026-07-05 problem sweep (transcript-mined)
Activity