Operational Safety, Traces, Replay, and Migrations for Any-Language Repositories

Purpose

Add the safety layer that lets a single-mainline, continuously deployed repository operate without relying on long-lived branches as a risk buffer.

The skill focuses on:

• production trace and log correlation
• replayable representative scenarios
• flag-aware testing
• deployment observability
• kill switches and rollback drills
• expand-contract data migrations
• compatibility checks across versions
• traceability from AI-authored changes to issue, PR, deploy, flag, and rollback evidence

Preconditions