mcp-skills

# mcp-skills Documentation Comprehensive documentation for mcp-skills - dynamic RAG-powered skills for code assistants via MCP. ## Quick Links - **[Main README](../README.md)** - Installation and quick start guide - **[Contributing Guidelines](../CONTRIBUTING.md)** - Development workflow and contribution guidelines - **[Project Summary](../PROJECT_SUMMARY.md)** - Project overview and objectives ## Documentation Structure ### Architecture **[Architecture Design](architecture/README.md)** - Comprehensive architecture documentation Technical design and implementation details: - System architecture and component design - Hybrid RAG approach (vector store + knowledge graph) - MCP server implementation - Toolchain detection system - Phase-by-phase implementation plan ### Skills Resources **[Repository Resources](skills/RESOURCES.md)** - Comprehensive skills repository index Complete catalog of skill repositories: - **Official Collections**: anthropics/skills (priority 1) - **Community Collections**: obra/superpowers, claude-mpm-skills, etc. - **Coding Skills**: Testing, debugging, refactoring, documentation - **Toolchain Skills**: Python, TypeScript, Rust, Go, Java - **Operations Skills**: CI/CD, cloud platforms, security, IaC - **MCP Servers**: Integration capabilities for skill-like functionality - **Integration Guide**: How to add repositories to mcp-skills - **Verification Status**: Tracked status of all repositories ### Research **[Skills Research](research/skills-research.md)** - Comprehensive skills ecosystem research Detailed research covering: - 69+ individual skills analyzed - Official Anthropic resources and MCP servers - Community collections and curated lists - 200+ MCP servers for tool integration - Language-specific skills (Python, TypeScript, Rust, Go, Java) - DevOps and operations skills - Quality indicators and selection criteria - Installation methods and licensing ## Documentation Index ### Getting Started 1. [Installation](../README.md#installation) - Install mcp-skills from PyPI or source 2. [Quick Start](../README.md#quick-start) - Setup wizard and basic usage 3. [Configuration](../README.md#configuration) - Global and project configuration ### Core Concepts 1. [Architecture Overview](architecture/README.md#system-architecture) - System design 2. [Hybrid RAG Approach](architecture/README.md#hybrid-rag-architecture) - Vector + KG search 3. [Toolchain Detection](architecture/README.md#toolchain-detection) - Auto-detection system 4. [Skill Discovery](architecture/README.md#skill-discovery-process) - How skills are found ### Skills Management 1. [Repository Index](skills/RESOURCES.md) - All available skill repositories 2. [Adding Repositories](skills/RESOURCES.md#integration-guide) - Add new skill sources 3. [Priority System](skills/RESOURCES.md#priority-system) - Repository precedence 4. [Verification Status](skills/RESOURCES.md#verification-status) - Repository validation ### Development 1. [Contributing](../CONTRIBUTING.md) - Development workflow 2. [Architecture Design](architecture/README.md) - Technical implementation 3. [Component Design](architecture/README.md#component-design) - Individual components ### Research & Background 1. [Skills Ecosystem Research](research/skills-research.md) - Comprehensive analysis 2. [Official Anthropic Resources](research/skills-research.md#official-anthropic-resources) - Official skills 3. [Community Collections](research/skills-research.md#community-collections-and-curated-lists) - Community skills 4. [MCP Servers](research/skills-research.md#mcp-servers-essential-tool-integrations) - Tool integrations ## Key Features ### Dynamic RAG-Powered Skills - **Vector Search**: Semantic similarity search over skill descriptions - **Knowledge Graph**: Skill relationships, dependencies, categories - **Hybrid Approach**: Combines vector + graph for better discovery - **On-Demand Loading**: Skills loaded when needed, not at startup ### Intelligent Recommendations - **Toolchain Detection**: Auto-detects Python, TypeScript, Rust, Go, Java, etc. - **Framework Detection**: Recognizes Flask, Django, React, Next.js, etc. - **Context-Aware**: Recommends skills based on project context - **Priority System**: Resolves conflicts via repository priorities ### Multi-Source Integration - **Git Repositories**: Clone and index skills from multiple sources - **Automatic Updates**: Configurable auto-update for repositories - **Priority Management**: Higher priority repositories override lower - **Conflict Resolution**: Smart handling of duplicate skills ### MCP Native - **First-Class MCP**: Built on Model Context Protocol - **Standard Protocol**: Works with any MCP-compatible code assistant - **Stdio Transport**: Standard input/output communication - **Resource Provider**: Exposes skills as MCP resources ## Common Tasks ### Installing mcp-skills ```bash pip install mcp-skills ``` ### Initial Setup ```bash mcp-skills setup ``` ### Starting the Server ```bash mcp-skills serve ``` ### Managing Repositories ```bash # Add a repository mcp-skills repo add https://github.com/anthropics/skills --priority 100 # List repositories mcp-skills repo list # Update all repositories mcp-skills repo update ``` ### Searching Skills ```bash # Search for skills mcp-skills search "testing" # Get skill details mcp-skills info pytest-skill # Get recommendations mcp-skills recommend ``` ## Technology Stack - **Python 3.11+**: Core implementation language - **MCP Python SDK**: Model Context Protocol integration - **ChromaDB/Qdrant**: Vector store for semantic search - **NetworkX/Neo4j**: Knowledge graph for relationships - **sentence-transformers**: Embedding generation - **SQLite**: Metadata and configuration storage - **GitPython**: Repository management ## Project Structure ``` mcp-skills/ ├── docs/ # Documentation │ ├── README.md # This file │ ├── architecture/ # Architecture design │ │ └── README.md # Detailed architecture │ ├── research/ # Research and analysis │ │ └── skills-research.md # Skills ecosystem research │ └── skills/ # Skills resources │ └── RESOURCES.md # Repository index ├── src/ # Source code │ └── mcp_skills/ # Main package ├── tests/ # Test suite ├── README.md # Main README ├── CONTRIBUTING.md # Contributing guidelines └── PROJECT_SUMMARY.md # Project summary ~/.mcp-skills/ # User data directory ├── config.yaml # User configuration ├── repos/ # Cloned skill repositories │ ├── anthropics-skills/ │ ├── obra-superpowers/ │ └── custom-repo/ ├── indices/ # Vector + KG indices │ ├── vector_store/ │ └── knowledge_graph/ └── metadata.db # SQLite metadata ``` ## Support - **Issue Tracker**: [GitHub Issues](https://github.com/bobmatnyc/mcp-skills/issues) - **Documentation**: [GitHub Wiki](https://github.com/bobmatnyc/mcp-skills/wiki) - **MCP Registry**: [Model Context Protocol](https://registry.modelcontextprotocol.io) ## Contributing We welcome contributions! Please see: - [CONTRIBUTING.md](../CONTRIBUTING.md) for development workflow - [Architecture Design](architecture/README.md) for technical details - [Skills Resources](skills/RESOURCES.md) to propose new repositories ## License MIT License - see [LICENSE](../LICENSE) for details --- **Last Updated**: 2025-11-21 **Status**: Early development - MVP in progress **Maintained by**: mcp-skills project