mcp-me

Your AI assistants don't know who you are. Every time you start a conversation with Claude, Copilot, Cursor, or Windsurf, it's a blank slate — no context about your skills, your projects, your career, or what you care about.

mcp-me fixes that. It creates a structured personal profile that any AI assistant can read via the Model Context Protocol (MCP). Think of it as a digital identity layer for AI — your bio, career, skills, interests, projects, and more, always available to every AI tool you use.

You: "Write me a cover letter for this job"
AI:  (reads your me://career, me://skills, me://projects)
     "Based on your 5 years at Acme Corp, your TypeScript expertise,
      and your open-source work on mcp-me..."

Why mcp-me?

• AI that knows you — Your assistants remember your skills, career, projects, and personality across every conversation
• Auto-generated — Pull data from 329 registered generators (implemented across 44 generator source files) with one command
• Privacy-first — All data stays local in YAML files on your machine. Nothing is sent to any cloud.
• Real-time plugins — 13 live integrations (Spotify now playing, GitHub repos, Last.fm scrobbles) that AI queries on demand
• Extensible — Community-driven generators and plugins. Add a new data source in ~10 lines of code.
• Works everywhere — Claude Desktop, Cursor, Windsurf, Copilot, and any MCP-compatible AI assistant

Installation

Prerequisite: Node.js 20 or later. Verify with node -v.

One-Click Install

Add mcp-me to your AI assistant in one click — no path configuration needed (profile defaults to ~/.mcp-me):

Claude Desktop: Download the latest .mcpb release and double-click to install, or drag it into Claude Desktop → Settings → Extensions.

After installing, initialize your profile:

mcp-me init
mcp-me generate --github your-username

npm (recommended)

npm install -g mcp-me

This makes the mcp-me command available everywhere on your system. No need to clone any repository — npm downloads the package for you.

mcp-me --help
mcp-me init
mcp-me generate --github your-username
mcp-me serve

Alternative — run without installing (via npx):

npx mcp-me --help

npx downloads the package temporarily and runs it. Useful for trying mcp-me once, but slower on repeated use since it re-downloads each time.

Profile location: Commands default to ~/.mcp-me. Override with MCP_ME_PROFILE_DIR or pass an explicit path: mcp-me serve ~/my-profile.

Quick Start

# 1. Initialize your profile (creates YAML templates + .mcp-me.yaml in ~/.mcp-me)
mcp-me init

# 2. Edit the config file — uncomment your sources
code ~/.mcp-me/.mcp-me.yaml

Your .mcp-me.yaml config file:

generators:
  github: your-username
  devto: your-username
  bluesky: handle.bsky.social
  zodiac: aquarius

plugins:
  github:
    enabled: true
    username: your-username

# 3. Generate! Reads sources from .mcp-me.yaml automatically
mcp-me generate

# 4. Start the MCP server
mcp-me serve

CLI flags also work: mcp-me generate --github octocat --devto myuser

All commands work with npx (zero install) or with mcp-me directly if installed globally. The generate command pulls your data from public APIs and auto-populates profile YAML files — no API keys needed for most sources.

Profile directory structure

~/.mcp-me/             → Default profile location
  .mcp-me.yaml        → Configuration (generators + plugins)
  identity.yaml        → Your data (name, bio, contact)
  skills.yaml          → Your data (languages, tools)
  projects.yaml        → Your data (portfolio)
  career.yaml          → Your data (experience)
  interests.yaml       → Your data (hobbies, topics)
  personality.yaml     → Your data (traits, values)
  goals.yaml           → Your data (short/long-term)
  faq.yaml             → Your data (Q&A pairs)

Configure Your AI Assistant

One-Click (Cursor & VS Code)

Use the one-click install badges above. After installing, run mcp-me init to create your profile.

Cursor Plugin (Open Plugins)

mcp-me ships as a full Open Plugins plugin for Cursor. Install from cursor.directory or clone the repo — the plugin auto-detects all components:

Validate the plugin structure: npm run validate:open-plugin

Windsurf

Add to ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "me": {
      "command": "npx",
      "args": ["-y", "mcp-me", "serve"]
    }
  }
}

Cursor (manual)

Add to .cursor/mcp.json in your project root:

{
  "mcpServers": {
    "me": {
      "command": "npx",
      "args": ["-y", "mcp-me", "serve"]
    }
  }
}

VS Code (GitHub Copilot)

Add to .vscode/mcp.json in your project root (or in your User Settings for global access):

{
  "servers": {
    "me": {
      "command": "npx",
      "args": ["-y", "mcp-me", "serve"]
    }
  }
}

Tip: To enable it globally (all workspaces), open VS Code Settings (Ctrl+Shift+P → "Preferences: Open User Settings (JSON)") and add the mcp key there instead.

Claude Desktop

Option A — Desktop Extension (.mcpb, recommended):

1. Download mcp-me.mcpb from GitHub Releases
2. Double-click the file, or drag it into Claude Desktop → Settings → Extensions
3. Set your profile directory when prompted (default: ~/.mcp-me)
4. Run mcp-me init if you haven't created a profile yet

Option B — Manual config:

Add to your Claude Desktop config:

{
  "mcpServers": {
    "me": {
      "command": "npx",
      "args": ["-y", "mcp-me", "serve"]
    }
  }
}

Agent Instruction Files

AI agents read special Markdown files in your project to understand how they should behave. Pair them with mcp-me so that every AI session automatically consults your profile — no more repeating your stack, career, or preferences in every conversation.

Quick setup:

# AGENTS.md — for Windsurf, Cline, Codex, etc.
cp "$(npm root -g)/mcp-me/templates/AGENTS.md" ./AGENTS.md

# GitHub Copilot
mkdir -p .github
cp "$(npm root -g)/mcp-me/templates/.github/copilot-instructions.md" .github/copilot-instructions.md

Both templates are included with mcp-me and tell the AI to call ask_about_me whenever context about you would improve the response.

See the Agent Instructions Guide for examples, global setup, and tips on combining with project-specific rules.

Profile Schema

Your profile is a collection of YAML files:

See Schema Reference for full documentation.

MCP Interface

Resources

Static profile data exposed as MCP resources:

• me://identity — Personal identity and contact
• me://career — Professional history
• me://skills — Skills and proficiencies
• me://interests — Hobbies and preferences
• me://personality — Personality traits and values
• me://goals — Personal and professional goals
• me://projects — Portfolio and projects
• me://faq — Frequently asked questions

Tools

• ask_about_me — Free-form question about the user
• search_profile — Keyword search across all profile data

Prompts