Agent Error Recovery Playbook

A comprehensive playbook covering 30+ failure scenarios agents encounter in production: API timeouts, auth token expiry, rate limiting, malformed responses, dependency conflicts, sandbox restrictions, file permission errors, and more. Each scenario includes a decision tree, retry strategy, fallback approach, and example recovery code. Built from real-world experience running Claude Code across hundreds of tasks. Includes ready-to-use error handling snippets in Python, TypeScript, and Bash.