op-recoverylisted

# op-recovery — diagnose, then act When the session is going wrong, the manual has a triage table and a set of moves. Diagnose the failure first, then apply the matching move. Don't skip the diagnosis — wrong move on right symptom is still the wrong move. ## Index | Question / situation | Atomic file | |---|---| | Triage: what's actually going wrong? Symptom → likely failure → first move | `~/.achron-spine/chapters/recovery/17a-failure-triage.md` | | The 7 recovery moves (restart, verify, shrink scope, read diff, roll back, refactor, browser-verify) + update-the-system + when-to-walk-away | `~/.achron-spine/chapters/recovery/17b-recovery-moves.md` | | High-stakes recoveries: leaked secret, mid-flight migration on prod, auth/RLS data leak, "scared to touch the codebase" | `~/.achron-spine/chapters/recovery/17c-high-stakes-cases.md` | ## How to use 1. **Always start with 17a** unless the situation is clearly one of the high-stakes cases in 17c. 2. From 17a's triage table, pick the matching first move, then jump to 17b for the full move. 3. **For secrets / prod migration / RLS leak / lost comprehension → 17c immediately.** These have a different order-of-operations from the generic moves. ## High-priority triggers - User says "are you sure?" twice → drift, restart with constraint stated upfront (17b Move A). - User says "you said the opposite earlier" → drift confirmed, fresh terminal (17b Move A). - Claude can't find a file/function it just referenced → hallucination che