MCP Skillset by Bobmatnyc is a standalone Python application that functions as a dynamic RAG-powered skills service for code assistants. It enhances AI agents by providing intelligent, context-aware skills through a hybrid RAG approach, combining vector and knowledge graph search. The service connects to multiple Git repositories for skill acquisition and utilizes ChromaDB for persistent local vector storage. It is compatible with any Model Context Protocol-enabled AI agent, offering secure, dynamically discoverable capabilities to developers and teams.

What does MCP Skillset connect to?

MCP Skillset connects to AI agents and code assistants using the Model Context Protocol. It acts as an MCP server tool, delivering intelligent, context-aware skills. The service employs a hybrid RAG approach, integrating vector and knowledge graph search capabilities. It acquires skills by connecting to multiple Git repositories and uses ChromaDB as a persistent local vector store. This architecture allows it to serve as a central hub for dynamic skill provision within the MCP ecosystem.

Core features and tools

MCP Skillset's core function is exposing dynamically discovered and loaded "skills" to AI clients. These skills are portable instruction sets or reusable task modules designed to enhance an AI agent's domain knowledge. Instead of fixed named tools, MCP Skillset enables runtime skill discovery and dynamic loading, presenting these skills as MCP prompts. It includes security features like prompt injection detection, multi-level threat classification, repository trust levels, size limits, and content sanitization.

Who is it for?

MCP Skillset targets developers and teams who utilize code assistants and AI agents requiring intelligent, context-aware, and dynamically discoverable skills. It benefits those needing to enhance their AI agents' capabilities with domain-specific knowledge and secure execution. The service is particularly useful for environments where AI agents need to adapt to varying project contexts and toolchains. Skip this if your AI agents do not require dynamic skill discovery based on project context.

How to get started

For remote servers using streamable HTTP, access is granted through a standard OAuth authorization flow. Local stdio servers may use environment variables or host process credentials.

Use cases

• Dynamic Skill Discovery and Loading: AI agents discover and load skills at runtime based on project context and toolchain.
• Intelligent Skill Recommendations: Automatically recommends relevant skills by detecting the project's toolchain (e.g., Python, TypeScript, Rust, Go).
• Secure Skill Execution: Enables safe execution of skills from various sources through multi-layer security validations.

MCP Skillset offers AI agents dynamic, context-aware skills by integrating hybrid RAG and connecting to Git repositories. Its security features, including prompt injection detection and content sanitization, ensure safe skill execution. The service supports various MCP-compatible AI clients, making it a valuable addition for developers and teams. Try enhancing your AI agent's Python project awareness using its intelligent skill recommendation feature.