Supercharge your AI coding workflow with standardized agent skills
A comprehensive collection of reusable Agent Skills for Claude Code, GitHub Copilot, Cursor, and other AI coding assistants. Install once, use everywhere.
The Skills Catalog is a curated collection of Agent Skills that teach AI coding assistants how to:
- π Manage backlogs and tasks systematically
- π Follow git hygiene and version control best practices
- ποΈ Refactor code safely and effectively
- β Implement test-driven development (TDD)
- π Set up CI/CD pipelines
- π Write excellent technical documentation
- π¬ Conduct systematic technical research
- π Orchestrate multi-agent workflows (tech leads)
Key Features:
- β Automated Setup β AI agents install everything autonomously
- β Works Everywhere β Claude Code, GitHub Copilot, Cursor, and more
- β Open Standard β Based on Agent Skills
- β Production Ready β Battle-tested across multiple real projects
- β Extensible β Create custom skills for your team
Let your AI agent install everything automatically:
Read and execute: https://raw.githubusercontent.com/All-The-Vibes/skills-catalog/main/AGENT_START.md
Read and execute: https://raw.githubusercontent.com/All-The-Vibes/skills-catalog/main/AGENT_START.md
Read and execute: https://raw.githubusercontent.com/All-The-Vibes/skills-catalog/main/AGENT_START.md
That's it! The agent will:
- β Detect your environment
- β Install bd (beads) for task management
- β Ask which skills you need
- β Download and configure everything
- β Verify the installation
Prefer to install manually? See INSTALLATION_GUIDE.md for detailed step-by-step instructions.
Quick manual install:
# 1. Install bd (beads)
curl -fsSL https://raw.githubusercontent.com/ezrasingh/beads/main/install.sh | sh
# 2. Initialize bd in your project
cd /path/to/your/project
bd init
# 3. Download skills
git clone https://github.com/All-The-Vibes/skills-catalog.git /tmp/skills-catalog
cp -r /tmp/skills-catalog/.claude/skills .claude/
cp /tmp/skills-catalog/AGENTS.md AGENTS.md
rm -rf /tmp/skills-catalog
# 4. Restart your editor/agentLocated in .claude/skills/core/:
| Skill | Purpose |
|---|---|
| bd | Task management with beads CLI |
| git | Version control hygiene for multi-agent work |
| subagent | Delegate complex tasks to specialized agents |
| technical-lead-role | Orchestrate multi-agent workflows (for tech leads) |
| task-delegation | Effective work delegation patterns |
| creating-skills | Learn to author new agent skills |
| session-continuation | Maintain context across sessions |
Located in .claude/skills/engineering/:
| Skill | Purpose |
|---|---|
| refactoring | Safe code restructuring with systematic methodology |
| tdd | Test-Driven Development practices and patterns |
Located in .claude/skills/devops/:
| Skill | Purpose |
|---|---|
| ci-cd | Continuous Integration/Deployment best practices |
Located in .claude/skills/documentation/:
| Skill | Purpose |
|---|---|
| technical-writing | Write clear, effective technical documentation |
Located in .claude/skills/research/:
| Skill | Purpose |
|---|---|
| technical-research | Systematic technical investigation and analysis |
Not sure which skills you need? See SKILL_SELECTOR.md for guidance.
Skills are markdown files that teach AI agents specialized capabilities:
- Agent reads AGENTS.md β Understands the workflow
- Skills activate automatically β Based on task type and role
- Agent follows skill guidance β Step-by-step instructions
- Work gets done right β Following best practices
Create a task:
bd create "Refactor authentication module"Tell your agent:
Use the refactoring skill to safely refactor the auth module
Agent executes following the skill:
- β Analyzes current code structure
- β Identifies refactoring opportunities
- β Makes changes incrementally with tests
- β Verifies functionality preserved
- β Commits with clear messages
| Document | Purpose |
|---|---|
| AGENT_START.md | Automated setup for AI agents |
| INSTALLATION_GUIDE.md | Manual installation instructions |
| SKILL_SELECTOR.md | Choose the right skills for your needs |
| VERIFICATION_CHECKLIST.md | Verify your installation |
| AGENTS.md | Workflow patterns and best practices |
- Getting Started: Use automated setup with your AI agent
- Choosing Skills: Read SKILL_SELECTOR.md
- Understanding Workflows: Read AGENTS.md
- Creating Skills: See
.claude/skills/core/creating-skills/SKILL.md - Contributing: Submit PRs with new or improved skills
- Standards: Follow the Agent Skills open standard
skills-catalog/
βββ README.md # You are here
βββ AGENT_START.md # Automated setup for AI agents
βββ INSTALLATION_GUIDE.md # Manual installation guide
βββ SKILL_SELECTOR.md # Help choosing skills
βββ VERIFICATION_CHECKLIST.md # Verify installation
βββ AGENTS.md # Workflow patterns
βββ .gitignore # Git ignore rules
βββ .claude/
βββ skills/
βββ core/ # Essential skills
β βββ bd/
β βββ git/
β βββ subagent/
β βββ technical-lead-role/
β βββ task-delegation/
β βββ creating-skills/
β βββ session-continuation/
βββ engineering/ # Software development
β βββ refactoring/
β βββ tdd/
βββ devops/ # Infrastructure & deployment
β βββ ci-cd/
βββ documentation/ # Writing & docs
β βββ technical-writing/
βββ research/ # Investigation & analysis
βββ technical-research/
The skills in this catalog follow these principles:
- State intent before acting
- Track all work in backlog
- No silent work
- Always pull before reading/editing
- Make decisions on current state
- When in doubt, pull
- One issue, one task
- Stay in scope
- Finish before starting new work
- Update issue status
- Commit frequently
- Report blockers immediately
- Push commits before stopping
- Sync backlog
- Never leave work stranded
- Test before presenting
- Fix obvious issues
- Verify it works
- Every line must justify its existence
- No stubs or placeholders
- Build working code or don't build it
Use the creating-skills skill as your guide:
# Read the creating-skills skill
cat .claude/skills/core/creating-skills/SKILL.md
# Or ask your agent
"Use the creating-skills skill to create a new [skill-name] skill"All skills follow the Agent Skills open standard
Skills are markdown files β edit them to fit your workflow:
# Edit any skill
vim .claude/skills/core/git/SKILL.md
# Commit changes
git add .claude/skills/core/git/SKILL.md
git commit -m "docs: customize git skill for our workflow"Simply delete skill directories you don't use:
# Remove skills not relevant to your project
rm -rf .claude/skills/research # If not doing research work- β Task management with bd
- β Git best practices
- β Refactoring guidance
- β TDD support
- β CI/CD setup
- β Multi-agent orchestration
- β Task delegation patterns
- β Team workflow coordination
- β Code review guidance
- β Documentation standards
- β Shared skill library
- β Consistent practices
- β Onboarding new members
- β Cross-project reuse
- β Custom skill creation
We welcome contributions!
- Use the
creating-skillsskill to create your skill - Place it in the appropriate category
- Follow the Agent Skills standard
- Test with real AI agents
- Submit a pull request
- Edit the
SKILL.mdfile - Keep the format consistent
- Add examples and clarity
- Test your changes
- Submit a pull request
Found a problem? Open an issue with:
- Skill name
- Issue description
- Expected vs actual behavior
- Steps to reproduce
MIT License β Use freely in your projects.
See LICENSE for details.
This catalog consolidates and standardizes skills developed and refined across multiple production projects, adapted into a general-purpose, standardized format for any software development workflow.
Q: How do skills activate? A: Automatically based on your role (tech lead vs individual contributor) and task description. Skills can also be explicitly referenced.
Q: Can I use this with GitHub Copilot / Cursor / other agents? A: Yes! Skills are standard markdown files readable by any AI coding assistant.
Q: Do I need to install all skills? A: No. Use SKILL_SELECTOR.md to choose what you need, or install everything and ignore what you don't use.
Q: Can I modify skills for my team? A: Absolutely! Skills are designed to be customized. Edit the SKILL.md files to fit your workflow.
Q: How do I create custom skills?
A: Use the creating-skills skill as your guide. It teaches the standard format and best practices.
Q: What's the difference between bd and Backlog MCP? A: Both manage backlogs. bd is a standalone CLI tool, Backlog MCP is an MCP server that integrates with editors. Choose based on your preference.
For AI-driven setup:
Read and execute: https://raw.githubusercontent.com/All-The-Vibes/skills-catalog/main/AGENT_START.md
For manual setup: See INSTALLATION_GUIDE.md
Need help choosing skills? See SKILL_SELECTOR.md
Last Updated: 2026-01-09 Version: 1.0 License: MIT