failure-handling

name	failure-handling
description	This skill should be used when handling failures in pipeline command flows — Mio blocking with NEEDS_CHANGES / BLOCKED, and infinite-loop protection for repeated same must-fix. Applies to /kon:quick, /kon:team, /kon:debug.

Failure Handling

Consumers: /kon:team, /kon:quick, /kon:debug.

Core principles (always): follow skills/core-principles — first principles (don't hide the issue); simplest, most concise correct solution.

Mio blocks (NEEDS_CHANGES / BLOCKED)

1. Do not forward Mio's output through the orchestrator — full review is in sessions/<SID>/review.md (subagentStop hook on team/quick/debug/review).
2. Resume 🎶 Yui (do not respawn with full agents/Yui.md if get-task-agent has an id) — delta prompt: read review.md, fix each must-fix by number.
3. Resume 🧹 Sawako when applicable — delta prompt only (files changed, no review body).
4. Resume 📝 Mio (do not respawn with full strict-review if id exists) — delta prompt: re-read review.md, verify each prior must-fix item-by-item.

See skills/teammate-flow Implementation loop — Task resume.

Infinite-loop protection

• Mio blocks the same must-fix 2 consecutive times → stop, ask the user (the plan itself may be the problem)

"Same must-fix" definition: Mio's must-fix item points to the same file + same function/section + same category of problem. Yui changing the wording or the approach but leaving the root problem in place still counts as the same item.