How to Use the CoFounder.im MCP Server with AI Agents

First, create a project on CoFounder.im and run the AI agents to analyze your startup idea. The MCP server exposes your project data and agent outputs to any MCP-compatible AI coding agent, so you need completed agent outputs to pull.

Go to cofounder.im/users/settings and generate an API token. This bearer token (starting with 'cfr_') authenticates your MCP connection. You'll add it to your AI agent's configuration file.

Add the CoFounder.im MCP server to your agent's configuration. For Claude Code, add to .mcp.json; for Cursor, add to .cursor/mcp.json; for Windsurf, add to ~/.codeium/windsurf/mcp_config.json. The server URL is https://mcp.cofounder.im/mcp with your token in the Authorization header.

Once connected, your AI agent can use the 'list_projects' tool to see all your CoFounder.im projects. This returns project names, statuses, descriptions, and IDs that you'll use to fetch specific build specs.

Use the 'get_build_spec' tool with a project ID to retrieve the complete build specification. This includes all agent outputs: tech stack, MVP plan, UI/UX design, implementation plan, market research, and more — everything your AI agent needs to start building.

Now your AI coding agent has full context about your validated startup idea. Ask it to build features based on the tech stack recommendations, implement the MVP plan, or follow the implementation steps. The build spec serves as your project's source of truth.