anti-panic-protocol
di noelistired
Web & Frontend DevelopmentSintesi
Handle tool failures under pressure with bounded retries, clean user communication, and safe escalation.
Descrizione
Cosa fa questa skill
Defines a strict error-handling workflow for agents: identify the failure, attempt two bounded fixes, then escalate with plain-language communication. Prevents retry loops, duplicate outbound messages, and raw error dumps reaching the user. Includes hard stops for destructive or security-boundary situations.
Enforces a two-attempt ceiling on retries and requires verified outcomes before reporting success, preventing runaway loops and false "done" confirmations that other ad-hoc error handling skips.
Quando usarla
- Recovering from a rate-limited publish without spamming retries
- Handling a calendar event creation error with date/timezone fixes
- Escalating a timed-out login flow to the user for OAuth confirmation
- Preventing duplicate posts when an external publish action fails
- Reporting a failed file write as pending rather than falsely done
Esempio di workflow
User asks: Post my announcement to the blog
- 1
Attempt to publish the post via the configured tool
- 2
Receive rate-limit rejection, identify failure in one sentence
- 3
Retry immediately as fix #1
- 4
Retry with a delayed slot as fix #2
- 5
Stop retrying, send escalation message: what failed, what was tried, what happens next
Dit faalde: publish werd afgewezen door rate-limit. Ik heb 1) direct retry en 2) delayed retry geprobeerd. Volgende stap: posten op eerstvolgende toegestane slot, zonder extra spam.
Installazione
clawhub install noelistired/anti-panic-protocolEsegui il comando CLI oppure incolla il prompt in Claude Code, Cursor, Codex o nell'agente di coding che usi.