Procore Mcp Server

What is Procore Mcp Server?

Procore Mcp Server is a Model Context Protocol server that exposes the full Procore REST API to AI assistants like Claude. Built with TypeScript, it converts Procore's OpenAPI spec into a compact catalog and auto-generates MCP tools for every API operation, working with Claude Desktop, Claude Code, and any MCP-compatible client.

How to use Procore Mcp Server?

Clone the repository, install dependencies with npm install, configure environment variables (PROCORE_CLIENT_ID, PROCORE_CLIENT_SECRET, PROCORE_COMPANY_ID), place Procore's OpenAPI spec file at specs/combined_OAS.json, run npm run build to generate the catalog, run npm run auth for OAuth authentication, then start the server with npm start. Configure it as an MCP server in Claude Desktop or Claude Code using the provided JSON configuration.

Key features of Procore Mcp Server

• Build-time parser converts Procore's 41MB OpenAPI spec into a compact catalog
• Auto-generates individual MCP tools for every API operation
• 7 meta-tools for endpoint discovery, search, details, and execution
• OAuth authentication with token refresh and local storage
• HTTP client with rate limiting, retries, and pagination
• Runtime catalog loading, search, and endpoint filtering

Use cases of Procore Mcp Server

• Construction engineers querying Procore project data via an AI assistant
• Automating Procore workflows like retrieving submittals, RFIs, or daily logs
• Building custom MCP-compatible tools that integrate with Procore's API

FAQ from Procore Mcp Server

What are the prerequisites to run Procore Mcp Server?

Node.js 18+, a Procore Developer Portal account, an OAuth app with Authorization Code grant type, and a redirect URI set to http://localhost.

How does authentication work?

Run npm run auth once to complete the OAuth flow in your browser. Tokens are saved to ~/.procore-mcp/tokens.json and auto-refresh when expired.

Where does the OpenAPI spec come from?

You must obtain Procore's OpenAPI spec file from Procore's API documentation and place it at specs/combined_OAS.json. It is not included in the repository due to its ~41MB size.

Which AI clients are supported?

The server works with Claude Desktop, Claude Code, and any MCP-compatible client.

What happens at build time?

Scripts parse the OpenAPI spec to produce a compact data/catalog.json and individual endpoint detail files, then generate a tools manifest with one named MCP tool per API operation.