repo-state-recovery

Repo State Recovery

Use this skill when the user cares about what is true now, not what chat folklore says was probably true before the interruption.

Procedure

1. Recover recent context
   • use available session/thread context if it exists
   • read repo handoff files, plans, and state artifacts
2. Inspect git state
   • git status --short --branch
   • recent commits
   • git diff --stat
3. Verify current health
   • run the repo's declared tests or gates
   • if the repo has no clear test entrypoint, say so explicitly
4. Audit local-only risk
   • inspect untracked files
   • inspect ignored files if anything looks suspicious
5. Reconstruct the workstream
   • identify the last meaningful increment
   • identify what is committed, what is local, and what is ambiguous

Output

Give the user:

• where work had reached
• current test or gate status
• committed vs local-only state
• likely next actions
• any recovery-risk findings

Pitfalls

• Do not rely only on git history.
• Do not rely only on prior conversation.
• Do not skip ignored-file checks when source files seem to be missing.
• Do not say the repo is probably fine if you can verify it.