Claude Code hooks lifecycle

All seven hook events Claude Code fires during a session, in the order they run.

• SessionStart — CLI launches, before any user input. Use to inject env or banners.
• UserPromptSubmit — each user message. Use to validate or block disallowed prompts.
• PreToolUse — before every tool call. Use to gate, log, or mutate arguments.
• PostToolUse — after every tool call. Use to audit results.
• Notification — when Claude wants the user's attention. Push to Slack or terminal.
• Stop — main agent turn ends. Cleanup, summary log.
• SubagentStop — subagent finishes. Aggregate child results.

Hooks live in ~/.claude/settings.json or .claude/settings.json. A return code of 2 blocks the action and surfaces stderr to Claude.

Related code examples

• Claude MCP Server Python Example — Build a Model Context Protocol Server
• Claude API Tool Use (Function Calling) Python Example
• Claude Agents Python — Agentic Loop with Tool Use (2026 Guide)