tally

This MCP server brings Tally.so form management capabilities into your AI workflows. With the Tally MCP Server, you can create, modify, retrieve, and manage forms and their submissions through natural language prompts from an MCP client. It supports operations such as creating forms with custom fields, updating form configurations, listing forms, retrieving form details, exporting submissions, and managing workspaces and teams. The server is designed to work with Claude Desktop, Cursor, Windsurf, and any MCP-compatible client via HTTP Stream transport, enabling a seamless, code-free integration into your assistant pipelines.

To use it, configure an MCP client to connect to the provided MCP endpoint (either via mcp-remote or direct HTTP Stream transport). You will authenticate using a Bearer token and authorize access with your Tally data through the TALLY_API_KEY. Once connected, you can issue natural-language requests like “Create a client intake form with fields name, email, and project details” or “Export submissions for the last 30 days,” and the MCP server will translate these into Tally API calls and return structured results in your chat or application interface.

Prerequisites:

• Node.js and npm installed on your machine
• Access to the internet
• A Tally API key (TALLY_API_KEY) and a server authentication token (AUTH_TOKEN) configured in Cloudflare Workers

Step-by-step:

1. Clone the repository (or install the MCP server package if published):

   • git clone <repository-url>
   • cd tally-mcp-server

2. Install dependencies:

   • npm install

3. Create or update a local environment config with required tokens (example using a .env file):

   • echo 'TALLY_API_KEY=your_tally_api_key' > .env
   • echo 'AUTH_TOKEN=your_auth_token' >> .env

4. Run the development server (for local testing):

   • npm run dev

5. If using the Cloudflare Workers deployment path, ensure your workers config imports TOKENs correctly and deploy per your usual workflow (wrangler or equivalent). After deployment, point MCP clients to the deployed endpoint and use the Bearer token authentication as described in the README.

6. Optional: Run tests and linting as defined by your project:

   • npm test
   • npm run test:coverage

Note: The exact deployment steps may vary based on whether you deploy to Cloudflare Workers directly or through a local proxy; adapt the environment variables and startup command accordingly.

Tips and common considerations:

• Ensure TALLY_API_KEY and AUTH_TOKEN are kept secure and not checked into version control.
• If you encounter transport issues, verify that your MCP client is configured to use HTTP Stream and that the Authorization header is correctly passed.
• For bulk operations, use the built-in safety features (preview → confirm → execute) in the MCP workflow to prevent accidental data loss.
• The server is designed to work with multiple MCP clients (Claude Desktop, Cursor, Windsurf, etc.). If you switch clients, ensure the transport method remains compatible (HTTP Stream recommended).
• Regularly update dependencies and keep test coverage above 90% to maintain reliability.