Claude Code Hooks for DevOps Workflows

Claude Code hooks let you run custom scripts before or after Claude executes tools. Configure them in ~/.claude/settings.json (user) or .claude/settings.json (project).

{
  "hooks": {
    "PostToolUse": [
      {
        "matcher": "Edit|Write",
        "hooks": [
          {
            "type": "command",
            "command": "npx prettier --write \"$(jq -r '.tool_input.file_path')\""
          }
        ]
      }
    ],
    "PreToolUse": [
      {
        "matcher": "Bash",
        "hooks": [
          {
            "type": "command",
            "command": "jq -r '.tool_input.command' >> ~/.claude/bash-log.txt"
          }
        ]
      }
    ]
  }
}

Available Hook Events

• PreToolUse - Before tool execution (can modify input, allow/deny)
• PostToolUse - After successful tool execution
• PostToolUseFailure - When a tool fails
• PermissionRequest - When permission dialog is shown
• UserPromptSubmit - When user submits a prompt (can inject context)
• Stop / SubagentStop - When agent finishes (can force continuation)
• SessionStart / SessionEnd - Session lifecycle (inject startup context)
• SubagentStart - When spawning a subagent
• Setup - During --init or --maintenance (repo initialization)
• PreCompact - Before context compaction (manual or auto)
• Notification - When notifications are sent

Hook Decision Control

Hooks can output JSON to control behavior. Exit code 0 with JSON stdout enables advanced control:

• PreToolUse: { "permissionDecision": "allow|deny|ask", "updatedInput": {...} }
• PostToolUse: { "decision": "block", "additionalContext": "..." }
• UserPromptSubmit: { "decision": "block", "reason": "..." }
• Stop: { "decision": "block", "reason": "keep working on X" }