mirror of
https://github.com/secondsky/claude-skills.git
synced 2026-09-18 19:54:22 +08:00
292c93845e
- Rewrites AI-sounding text to read human; 8 editing passes + review-only mode - Adapted from jpeggdev/humanize-writing (MIT); npm installer scaffolding dropped - Frontmatter condensed to 238 chars (sync enforces 250 max for single-skill plugins) - Lockstep bump 3.8.0 -> 3.9.0: all 145 plugins, both marketplaces, package.json - CHANGELOG entry for 3.9.0
Claude Hook Writer Skill
Production-ready skill for writing secure, reliable, and performant Claude Code hooks.
Auto-Trigger Keywords
- Claude Code hooks
- hook writer
- PreToolUse hook
- PostToolUse hook
- UserPromptSubmit hook
- SessionStart hook
- hook security
- hook validation
- PRPM package
- bash hook
- hook configuration
- hook testing
When to Use
- Creating new Claude Code hooks
- Reviewing hook code for security issues
- Debugging hook failures
- Optimizing slow hooks
- Publishing hooks as PRPM packages
- Understanding hook exit codes and behavior
Key Features
- Security Requirements: Input validation, path sanitization, sensitive file protection
- Reliability Patterns: Handle missing dependencies, set timeouts, log properly
- Performance Guidelines: Keep hooks fast, use specific matchers, dedupe operations
- Code Templates: Format on save, block sensitive files, command logger, security hooks
- Testing Strategies: Manual testing, edge cases, integration testing
- PRPM Publishing: Package structure, configuration, advanced options
Common Pitfalls Covered
- Not quoting variables (breaks on spaces)
- Trusting input without validation
- Blocking operations too long
- Using wrong exit codes
- Logging to stdout instead of stderr
- Assuming tools exist
Source
Adapted from PRPM claude-hook-writer
License
MIT