name: error-recovery description: Session resumption after unexpected CLI closures. Activated automatically when .claude-state.md exists. Restores original prompt, classification, sequence and progress position. Skill-agnostic — only cares about sequence index and basic context. priority: highest (pre-sequence execution) triggers: always (via classifier if state file detected)

Error Recovery – Crash-Resistant Session Resume

You are the persistence & resumption layer.

Goal: Let user continue exactly where they left off after CLI crash / restart.

Rules

• File: .claude-state.md (root level, gitignored)
• Never modify other skills' behavior
• Only read/write/delete this one file
• Minimal content: prompt + class + sequence + progress pointer + key side-effects
• On resume: inject recovered state, jump to next pending step
• Cleanup: delete file on successful commit-orchestrator or explicit /clearstate

Process (strict)

1. Classifier already detected file → CLASS: RESUME → you are first step
2. Read ".claude-state.md"
3. Parse sections (fail gracefully if malformed)
4. Show user summary of recovered state
5. Ask: Resume this session? [y/n]
   • y → output instruction to continue from next step
   • n → delete ".claude-state.md" → reply "State cleared. Normal classification will run next."
6. If resuming:
   • Inject recovered prompt + classification + sequence as context
   • Tell main loop: "Continue sequence from step N: <tool name> <arguments if any>"

Output Format – On detection

Detected resume file: .claude-state.md

Recovered state:

• Timestamp: <date>
• Original prompt: <first few lines or summary>
• Classification: <class>
• Sequence length: <n> steps
• Last completed: step <k> (<tool name>)
• Next step: step <k+1> (<tool name>)
• Branch: <branch or none>
• Modified files: <list or none>
• Pending approval: <yes/no – last question if yes>

Resume from here? [y/n]

(After user answer)

If y: Continuing from step <k+1>: <tool> <args if known>

If n: State file deleted. Re-classify normally.