What problem does it solve?
Claude Code hooks often fail silently: scripts that are not executable, blocking hooks that use exit 1 instead of exit 2, missing timeouts that hang the agent, or matchers attached to events that ignore them. This Skill systematically audits every hook across all settings files and walks you through fixing each one interactively.
Core Features & Use Cases
- Full hook discovery and scoring: Scans global and project settings.json files, parses every hook across all 30 event types, and scores each on a 10-point rubric covering event validity, matcher scope, command resolution, timeout, and blocking awareness.
- Filesystem and semantics validation: Resolves each command against the filesystem, checks executability, verifies exit-code strategy in blocking hook scripts, and flags interactive commands that would hang the agent.
- Interactive per-hook review: Walks through hooks one by one, confirms scope and usefulness with the user, applies fixes via Edit, and produces a final audit report with per-hook scores and a change summary.
- Use Case: After copying hooks from another machine, run the audit to discover that a PreToolUse script lacks execute permissions and uses exit 1, then fix both issues in the same session.
Quick Start
Ask the assistant to audit all Claude Code hooks in your settings files and fix any that are broken, misconfigured, or missing timeouts.
