Run It Twice v0.1.0

It worked once. Run it twice.

Run It Twice is a portable Agent Skill for testing whether one logical action still creates one real outcome across retries, duplicate delivery, concurrency, restarts, and crash windows.

Highlights:

• one focused 133-line skill with no runtime dependencies
• intent-versus-attempt identity, semantic payload binding, and authoritative effect counting
• safe replay matrix for retries, webhooks, jobs, payments, migrations, and recovery paths
• four evidence-scoped verdicts: REPEAT-SAFE, DUPLICATES, DIVERGES, and NOT TESTED
• reproducible 16-case paired benchmark with raw outputs and integrity checks
• complete English, Chinese, Portuguese, Japanese, Spanish, and Russian documentation

Install:
npx skills add Pablo-aps/run-it-twice

The benchmark is a single-run synthetic artifact-review snapshot, not a proof of exactly-once execution or a universal model-quality claim.