Agent Skills Collection
A collection of Agent Skills following the Agent Skills specification. These skills provide specialized capabilities for AI agents like Claude Code, OpenCode, Cursor, and other compatible platforms.
Overview
Agent Skills are folders of instructions, scripts, and resources that agents can discover and use to perform tasks more accurately and efficiently. This repository contains curated skills for common development workflows.
Available Skills
1. conventional-commits
Follow the Conventional Commits specification for writing standardized commit messages.
Features:
- Structured commit message format
- Semantic versioning integration
- Automated changelog generation
- Pre-commit hook integration
Use when: Creating git commits, writing commit messages, or working with semantic versioning.
2. skill-authoring
Create, edit, audit, and validate Agent Skills following TDD and specification compliance.
Features:
- TDD approach to skill creation (RED-GREEN-REFACTOR)
- Specification compliance validation
- Duplicate detection and resolution
- Best practices for documentation and token efficiency
Use when: Creating new skills, auditing existing ones, or ensuring specification compliance.
3. deep-wiki
Generate comprehensive wiki documentation for any codebase, similar to deepwiki.com.
Features:
- Code structure analysis
- Architecture documentation
- API reference generation
- Interactive navigation
- Multiple output formats (HTML, Markdown)
Use when: Documenting a codebase, creating wiki pages, or generating architecture documentation.
4. system-performance
Apply Brendan Gregg's performance analysis methodologies for investigating system performance issues.
Features:
- USE Method (Utilization, Saturation, Errors)
- RED Method (Rate, Errors, Duration)
- Flame graphs visualization
- Performance analysis workflows
- Linux performance tools reference
Use when: Analyzing system performance, debugging performance issues, or investigating bottlenecks.
5. code-reviewer
Conduct thorough code reviews with systematic checks for code quality, bugs, security issues, performance problems, and adherence to best practices.
Features:
- Systematic review process (6 phases)
- Security, quality, performance, testing review
- Code review checklist and feedback guidelines
- Before/after code comparisons
- Issue prioritization
Use when: Reviewing pull requests, conducting code reviews, or analyzing code quality.
6. security-expert
Identify and prevent security vulnerabilities using OWASP Top 10, secure coding practices, and systematic vulnerability analysis.
Features:
- OWASP Top 10 coverage
- Input validation and sanitization
- Authentication and authorization patterns
- API security best practices
- Dependency vulnerability scanning
Use when: Analyzing security vulnerabilities, conducting security reviews, or checking for common vulnerabilities.
7. frontend-design
Design and implement distinctive, production-ready frontend interfaces with strong aesthetic direction.
Features:
- Design thinking + execution approach
- Purpose-driven visual choices
- Support for HTML/CSS/JS, React, Vue, and more
- Accessibility and performance considerations
Use when: Creating or restyling web pages, components, dashboards, or application UIs.
8. native-web-search
Trigger native web search for fast internet research with concise summaries and full source URLs.
Features:
- Fast model with native web search
- Concise research summaries
- Explicit full source URLs
- Command-line interface
Use when: Performing internet research, finding current information, or getting quick answers with sources.
9. summarize
Fetch URLs or convert local files (PDF/DOCX/HTML/etc.) into Markdown for analysis, optionally with summarization.
Features:
- Convert web pages to Markdown
- Binary document conversion (PDF/DOCX/PPTX)
- Optional summarization
- Supports various file formats
Use when: Converting documents for analysis, pulling down web pages as Markdown, or summarizing long documents.
10. uv
Use uv instead of pip/python/venv for Python package and environment management.
Features:
- Fast Python package installer
- Inline script dependencies
- Virtual environment management
- Script execution with
uv run
Use when: Running Python scripts, managing dependencies, or creating reproducible Python environments.
11. caveman
Ultra-compressed communication mode to save tokens while keeping technical accuracy.
Features:
- Compressed communication style
- Persistence across turns
- Rules for dropping articles and filler
- Auto-clarity exceptions for critical info
Use when: Saving tokens, being brief, or when requested for "caveman mode".
12. grill
