attio

This MCP server provides full Attio CRM surface coverage, enabling AI assistants like Claude or ChatGPT to interact with your Attio workspace through natural language. It exposes universal tools to manage Companies, People, Deals, Tasks, Lists, and Notes, plus batch operations, content search, and relationship navigation. With the Attio MCP, you can ask the AI to search for records, create or update entities, link records, and perform complex workflows (e.g., multi-step pipelines or advanced filtering) across your entire Attio workspace. The built-in tools are designed to be generic and reusable, so you can compose natural-language prompts into robust CRM actions without relying on raw API calls.

To use it, enable the MCP under the attio server entry (attio) and point your AI integration to the MCP endpoint. The system supports advanced capabilities like universal search across records, relationship discovery between Companies and People, batch processing for bulk updates, and content search across notes, tasks, and lists. For developers, there are pre-built Claude skills and prompts to facilitate correct field names and workflows, plus setup guides for leveraging ChatGPT Developer Mode and Cloudflare integration if you’re hosting externally.

Prerequisites:

• Node.js version 20.x or newer
• npm (comes with Node.js)

Steps:

1. Install the MCP package globally (or use npx to run directly):

npm install -g attio-mcp

2. Create a configuration file (example: mcp_config.json) with the following content:

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

3. Start the MCP server using the configuration. If the package exposes a CLI, run something like:

attio-mcp --config mcp_config.json

4. Verify the server is running and the attio MCP endpoint is reachable. Configure your AI assistant to point to the MCP endpoint exposed by this server.

Optional: If you prefer to run via npx directly without global install, you can start with:

npx -y attio-mcp --config mcp_config.json

Note: The exact CLI flags may vary slightly depending on package version. Adjust the --config path and any environment-specific options as needed.

Environment variables and configuration tips:

• You can add environment variables for secrets or feature toggles as needed (example placeholders: "ENV_VAR_NAME": "description").
• When running behind proxies or in Cloudflare setups, ensure network access to Attio API endpoints.
• If you encounter field-name errors, consider using the built-in Claude skills (attio-workspace-schema, attio-skill-generator) to generate and validate your workspace schema.
• For security, enable the built-in approval flows for writes, and configure OAuth where applicable when deploying with ChatGPT Developer Mode.
• If you see memory or performance warnings, the MCP demonstrates high-efficiency, but you can adjust worker/container resources accordingly.
• Refer to the docs in the repository for migration guides and Cloudflare Worker deployment steps if you plan to host externally.