Complete reference for every Claude Code CLI flag and pattern you'll need. Covers the basics like interactive vs headless mode, but the real value is in the headless agent patterns. If you're spawning agents in CI/CD or automation, the session control and permission flags here are critical to avoid hangs. Shows exactly which flags to combine for headless execution, how to define custom subagents inline with JSON, and the streaming output formats for real-time monitoring. The examples are practical: piping logs for analysis, multi-turn sessions with resume, structured output validation. Keep this open when you're writing scripts that call Claude programmatically.
npx skills add https://github.com/parcadei/continuous-claude-v3 --skill cli-reference