mcp-me is a local MCP (Model Context Protocol) server designed to solve the problem of AI assistants lacking the user's personal background. It creates a structured personal profile for any MCP-compatible AI assistant to read, effectively providing a layer of digital identity for the AI so it understands the user's skills, career, interests, projects, and other information.
Purpose Every time a conversation starts with an AI assistant such as Claude, Copilot, Cursor, or Windsurf, the AI is in a context-free state, knowing nothing about the user's skills, projects, career, or focus areas. mcp-me generates a structured personal profile so that the AI can automatically obtain this information in every conversation, reducing the need to repeatedly describe one's own background.
Main capabilities
- Automatic profile generation: with a single command, it pulls data from public APIs via 329 registered generators (distributed across 44 generator source files and 15 batch files) and automatically fills in the YAML profile file.
- Privacy first: all data stays in a local YAML file and is not sent to any cloud.
- Real-time plugins: 13 live integrations are provided (such as Spotify now playing, GitHub repositories, Last.fm scrobbles), which the AI can query on demand.
- Extensible: community-driven generators and plugins; adding a new data source takes about 10 lines of code.
- Broad compatibility: supports Claude Desktop, Cursor, Windsurf, Copilot, and any MCP-compatible AI assistant.
How to use Prerequisite: Node.js 20 or higher; verify with node -v. One-click install: Claude Desktop users can download the latest .mcpb release and double-click to install, or drag it into Claude Desktop → Settings → Extensions. After installation, run mcp-me init to initialize the profile. Install via npm (recommended): npm makes the mcp-me command globally available without cloning the repository. npx can also be used to run it temporarily. Profile location: by default located at ~/.mcp-me; can be overridden via MCP_ME_PROFILE_DIR or an explicit path (such as mcp-me serve ~/my-profile). The configuration file is .mcp-me.yaml. ClThe I flag is likewise available, for example mcp-me generate --github octocat --devto myuser. Configuring AI assistants: supports Cursor (one-click install or manual configuration of .cursor/mcp.json), VS Code (.vscode/mcp.json), Windsurf (~/.codeium/windsurf/mcp_config.json), and Claude Desktop (desktop extension or manual configuration). Agent instruction files: AI agents read special Markdown files in the project (such as AGENTS.md or .github/copilot-instructions.md), which can be used with mcp-me so the AI automatically references the personal profile in every session.
MCP resources and tools Static profile data is exposed as MCP resources: me://identity, me://career, me://skills, me://interests, me://personality, me://goals, me://projects, me://faq. Tools include: ask_about_me (free-form questions), search_profile (keyword search), introduce_me (generates a two-paragraph introduction), summarize_career (summarizes career trajectory), technical_profile (describes technical skills and stack), collaboration_fit (assesses project fit).
Generators and plugins Generators run during mcp-me generate, automatically filling in the profile from public APIs or user-provided export files; most data sources require no API key. Some generator examples: --blogger-backup (imports a Blogger XML export), --letterboxd, --goodreads, --chess, --lichess, --codewars, --leetcode, --lastfm, --steam, --twitch, --dribbble, --unsplash, --exercism, --hackerrank, --anilist, --gravatar, --keybaetc. The plugin runs during mcp-me serve, providing real-time data for AI assistants. It can be enabled in .mcp-me.yaml, and community plugins are installed from npm and automatically discovered.
Use cases Suitable for users who want AI assistants to understand their own background during conversations, including developers, designers, writers, etc. It can be used to help AI provide more accurate answers related to personal skills, occupation, and projects, reducing repeated explanations.
Limitations The current materials do not provide specific limitation information beyond pricing, cost, and platform support; none is currently provided.
