Context Passport

A personal context layer MCP server. Fill in your profile once, then any MCP-compatible AI (Claude Desktop, Cursor, etc.) instantly knows who you are, what you're working on, and how you like to work — without you re-explaining yourself every session.

Everything is stored locally in context.json. Nothing leaves your machine (except when you explicitly call sync_from_github, which reads from the GitHub API).

Setup

1. Install

# Option A — clone and install
git clone https://github.com/karanhanda/context-passport
cd context-passport
npm install

# Option B — run directly with npx (no clone needed)
npx context-passport

2. Fill in your context

Open context.json and fill in your details:

{
  "name": "Jane Smith",
  "role": "Senior Product Manager",
  "working_on": ["Q2 roadmap", "mobile redesign"],
  "expertise": ["product strategy", "SQL", "user research"],
  "work_style": "Prefer concise answers. Always give me the 'why' before the 'how'.",
  "goals": "Ship mobile v2 by end of Q2, get promoted to Director by EOY",
  "do_not": ["don't give me bullet points for everything", "don't be sycophantic"],
  "current_project": {
    "name": "Mobile Redesign v2",
    "description": "Full redesign of the iOS and Android apps for 2M+ users",
    "status": "In design review",
    "next_steps": ["Finalize nav patterns", "Handoff to eng by May 1"]
  },
  "last_updated": ""
}

Or skip this entirely and call sync_from_github to auto-populate from your GitHub activity.

3. Connect to Claude Desktop

Open your Claude Desktop config file (locations below) and add the mcpServers block:

{
  "mcpServers": {
    "context-passport": {
      "command": "/usr/local/bin/node",
      "args": ["/ABSOLUTE/PATH/TO/context-passport/server.js"],
      "env": {
        "GITHUB_TOKEN": "ghp_your_token_here"
      }
    }
  }
}

The env block is optional — only needed for GitHub sync. Leave it out if filling in context.json manually.

Config file locations

| Platform | Path | |----------|------| | Mac | ~/Library/Application Support/Claude/claude_desktop_config.json | | Windows | %APPDATA%\Claude\claude_desktop_config.json |

After saving, restart Claude Desktop for the changes to take effect.

GitHub auto-sync

Instead of filling in context.json by hand, call sync_from_github and Context Passport will automatically pull in:

• Your recently active repos → working_on
• Languages used across those repos → merged into expertise
• Your open pull requests → stored in github.open_prs
• Your GitHub display name → name (if not already set)

Getting a GitHub token

1. Go to github.com/settings/tokens
2. Click Generate new token (classic)
3. Name it context-passport
4. Check the repo scope (or public_repo for public repos only)
5. Copy the token and paste it into env.GITHUB_TOKEN in your MCP config

How to test it

Once Claude Desktop is restarted, open a new conversation and type:

Load my context passport.

Claude will call get_context and greet you with a summary of your profile. Also try:

Sync my GitHub activity into my context.

What am I currently working on?

Update my context — I just started a new project called "API overhaul".

The four tools

| Tool | What it does | |------|-------------| | get_context | Returns your full profile as formatted markdown — name, role, expertise, goals, work style, and current project | | update_context | Updates any field(s) and saves to disk. Pass only the fields you want to change. | | get_current_project | Returns just your current project block. If empty, prompts the AI to ask you. | | sync_from_github | Pulls your repos, languages, and open PRs from GitHub and merges them into your profile. Requires GITHUB_TOKEN. |

Publishing / using via npx

Anyone can run Context Passport without cloning:

npx context-passport

Or install globally:

npm install -g context-passport

Then use context-passport as the command in the MCP config instead of a full node /path/to/server.js.

Listing on Smithery

Smithery is the public registry for MCP servers. To list Context Passport:

1. Push this repo to GitHub
2. Go to smithery.ai and sign in with GitHub
3. Click Submit a server and paste your GitHub repo URL
4. Smithery reads smithery.yaml automatically — config schema and install instructions are already set up

Privacy

All data is stored in context.json on your local machine. The server runs as a local process communicating only over stdin/stdout with your AI client. No data is sent to any external server. The only outbound network call is to api.github.com when you explicitly call sync_from_github.