AI Company MCP Server

AI Knowledge & Skills Server — Expose your entire AI knowledge base to any MCP client.

A production-ready Model Context Protocol (MCP) server that acts as your AI knowledge and skills hub. It automatically discovers, indexes, and serves your skill library to any MCP-compatible client.

✨ Features

• 🔍 Automatic Skill Discovery — Drop folders into knowledge/skills/ and they're instantly available

• 🚀 Full-Text Search — TF-IDF powered search with relevance ranking and filtering

• 📋 Rich Metadata — YAML frontmatter parsing with automatic inference

• 🔄 Hot Reload — File watcher automatically reindexes on changes

• 🛠 17 MCP Tools — List, search, get, recommend, validate, export, and more

• 📚 10 MCP Prompts — Frontend, backend, React, testing, architecture, and more

• 📦 MCP Resources — Browse skills, metadata, categories, tags via resource API

• ⚡ High Performance — In-memory index, caching, lazy loading

• 🌍 Cross-Platform — Windows, Linux, macOS

• 🔒 Runs Locally — No network, no external services, your data stays private

Related MCP server: skillsmp-mcp-server

🚀 Quick Start

# Run directly (no installation)
npx ai-company-mcp

Or install globally:

npm install -g ai-company-mcp
ai-company-mcp

📁 Skill Structure

Create skill directories in knowledge/skills/:

knowledge/skills/
└── react-patterns/
    ├── SKILL.md          # Primary file
    └── examples.md       # Additional files

Each skill's markdown file supports YAML frontmatter:

---
name: react-patterns
description: Common React patterns and best practices
category: frontend
tags: [react, patterns, hooks]
priority: high
difficulty: intermediate
---

🔌 Client Configuration

OpenCode

{
  "mcp": {
    "ai-company": {
      "command": "npx",
      "args": ["-y", "ai-company-mcp"]
    }
  }
}

Claude Desktop

{
  "mcpServers": {
    "ai-company": {
      "command": "npx",
      "args": ["-y", "ai-company-mcp"]
    }
  }
}

Cursor / VS Code / Windsurf

See examples/ for all client configurations.

🛠 Available Tools

⚙️ Configuration

# Environment variables
MCP_SKILLS_PATH=./my-skills
MCP_WATCH_ENABLED=true
MCP_LOG_LEVEL=debug

# CLI arguments
npx ai-company-mcp --skills-path=./my-skills --watch=false

# JSON config file (mcp-config.json)
{
  "skillsPath": "./my-skills",
  "watchEnabled": true,
  "logLevel": "debug"
}

See docs/configuration.md for full options.

📖 Documentation

• Usage Guide

• Configuration

• Architecture

• Development

• Troubleshooting

🧪 Development

git clone <repo-url>
cd ai-company-mcp
npm install
npm run dev       # Hot reload development mode
npm test          # Run tests
npm run build     # Production build

📄 License

MIT — see LICENSE for details.