mcpbeat

Debug Hooks

parcadei/debug-hooks

Systematic hook debugging workflow. Use when hooks aren't firing, producing wrong output, or behaving unexpectedly.

863 tokens
context cost
the whole folder, loaded on every use
1
files
instructions only
0
copies elsewhere
how many repositories repackaged it
3880
stars on the repo
on the repository, not the skill itself

Install

one command, takes just this skill from the repository
npx skills add https://github.com/parcadei/Continuous-Claude-v3 --skill debug-hooks

The instruction itself

12 sections, as written by the author

Debug Hooks

Systematic workflow for debugging Claude Code hooks.

When to Use

  • "Hook isn't firing"
  • "Hook produces wrong output"
  • "SessionEnd not working"
  • "PostToolUse hook not triggering"
  • "Why didn't my hook run?"

Workflow

1. Check Outputs First (Observe Before Editing)

# Check project cache
ls -la $CLAUDE_PROJECT_DIR/.claude/cache/

# Check specific outputs
ls -la $CLAUDE_PROJECT_DIR/.claude/cache/learnings/

# Check for debug logs
tail $CLAUDE_PROJECT_DIR/.claude/cache/*.log 2>/dev/null

# Also check global (common mistake: wrong path)
ls -la ~/.claude/cache/ 2>/dev/null

2. Verify Hook Registration

# Project settings
cat $CLAUDE_PROJECT_DIR/.claude/settings.json | grep -A 20 '"SessionEnd"\|"PostToolUse"\|"UserPromptSubmit"'

# Global settings (hooks merge from both)
cat ~/.claude/settings.json | grep -A 20 '"SessionEnd"\|"PostToolUse"\|"UserPromptSubmit"'

3. Check Hook Files Exist

# Shell wrappers
ls -la $CLAUDE_PROJECT_DIR/.claude/hooks/*.sh

# Compiled bundles (if using TypeScript)
ls -la $CLAUDE_PROJECT_DIR/.claude/hooks/dist/*.mjs

4. Test Hook Manually

# SessionEnd hook
echo '{"session_id": "test-123", "reason": "clear", "transcript_path": "/tmp/test"}' | \
  $CLAUDE_PROJECT_DIR/.claude/hooks/session-end-cleanup.sh

# PostToolUse hook (Write tool example)
echo '{"tool_name": "Write", "tool_input": {"file_path": "test.md"}, "session_id": "test-123"}' | \
  $CLAUDE_PROJECT_DIR/.claude/hooks/handoff-index.sh

5. Check for Silent Failures

If using detached spawn with stdio: 'ignore':

// This pattern hides errors!
spawn(cmd, args, { detached: true, stdio: 'ignore' })

Fix: Add temporary logging:

const logFile = fs.openSync('.claude/cache/debug.log', 'a');
spawn(cmd, args, {
  detached: true,
  stdio: ['ignore', logFile, logFile]  // capture stdout/stderr
});

6. Rebuild After Edits

If you edited TypeScript source, you MUST rebuild:

cd $CLAUDE_PROJECT_DIR/.claude/hooks
npx esbuild src/session-end-cleanup.ts \
  --bundle --platform=node --format=esm \
  --outfile=dist/session-end-cleanup.mjs

Source edits alone don't take effect - the shell wrapper runs the bundled .mjs.

Common Issues

| Symptom | Likely Cause | Fix |

|---------|--------------|-----|

| Hook never runs | Not registered in settings.json | Add to correct event in settings |

| Hook runs but no output | Detached spawn hiding errors | Add logging, check manually |

| Wrong session ID | Using "most recent" query | Pass ID explicitly |

| Works locally, not in CI | Missing dependencies | Check npx/node availability |

| Runs twice | Registered in both global + project | Remove duplicate |

Debug Checklist

  • [ ] Outputs exist? (ls -la .claude/cache/)
  • [ ] Registered? (grep -A10 '"hooks"' .claude/settings.json)
  • [ ] Files exist? (ls .claude/hooks/*.sh)
  • [ ] Bundle current? (ls -la .claude/hooks/dist/)
  • [ ] Manual test works? (echo '{}' | ./hook.sh)
  • [ ] No silent failures? (check for stdio: 'ignore')

Source Sessions

Derived from 10 sessions (83% of all learnings):

  • a541f08a, 1c21e6c8, 6a9f2d7a, a8bd5cea, 2ca1a178, 657ce0b2, 3998f3a2, 2a829f12, 0b46cfd7, 862f6e2c

How to use it

Copy the folder

Take parcadei/debug-hooks from the repository into ~/.claude/skills for personal use, or into .claude/skills inside a project.

Check the name does not clash

The agent identifies a skill by the name field in its header. Two skills with the same name cannot sit side by side — one of them will be ignored.

Install what it needs

The instructions reference npx. Without those the skill loads but fails at the first command.