# contactapi [Health: Active]

**Category:** 👤 Customer Data Platforms  
**Repository:** https://github.com/kozmos-tech/contactapi  
**GitHub Stars:** 1  
**Views:** 1  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/contactapi

## Description
MCP bridge for Contact API — save and manage contacts from your AI tools.

## Claude Desktop Quick Installation
Remote MCP endpoint (confidence: high). Install path detected from listing signals. Add as a URL/SSE server in your client:

```json
"mcpServers": {
  "contactapi": {
    "url": "https://contactapi.dev/mcp"
  }
}
```

## Documentation

## What contactapi MCP server does

The contactapi MCP server exposes Contact API contact records to supported MCP clients. It provides five tools that correspond to the service's REST operations: `list_contacts`, `get_contact`, `create_contact`, `update_contact`, and `delete_contact`.

Contact API stores contacts as JSON records that include an email and any additional fields supplied by the caller. The email acts as the record key for creation: if a contact already exists with that email, `create_contact` updates the existing record instead of creating a duplicate. The available operations cover the full contact lifecycle, including paginated listing, retrieving one record, editing a record, and removing a record.

## How it works

Connect an MCP-compatible client to `https://contactapi.dev/mcp`. Authentication uses OAuth 2.1 with dynamic client registration. The client discovers the authorization server from the MCP endpoint, registers itself, and opens a login and approval flow. After authorization, the tools operate within the account used during login.

The MCP layer mirrors Contact API rather than introducing a separate contact data model. Its five tools map to the service's create-or-upsert, list, fetch, update, and delete behavior. The server is implemented with Hono and deployed on Vercel, while the MCP endpoint and tool wiring are part of the application source.

## Setup and configuration

To use the contactapi MCP server, configure an MCP client with the remote endpoint `https://contactapi.dev/mcp`. The README specifically identifies Claude Desktop, claude.ai custom connectors, Cursor, and the MCP Inspector as clients that can connect to it. The client handles OAuth discovery, dynamic registration, sign-in, and consent; the documented flow does not require copying an API key into the MCP client.

For local application development, the repository lists the Vercel CLI as a prerequisite. Install dependencies with `npm install`, start a local development server with `vc dev`, and open `http://localhost:3000`. The project also documents `vc build` for building and `vc deploy` for deployment. These commands describe developing or deploying the application; the hosted MCP endpoint is the documented connection target for clients.

The underlying REST API accepts Bearer API keys. Secret keys have full endpoint access, while publishable keys can only create contacts and are restricted to configured domains. That key distinction applies to direct API use; the MCP connection uses the documented OAuth flow and scopes actions to the account that approves access.

_Full upstream README: https://allmcps.com/mcp/contactapi/readme_