This MCP server exposes the Discogs API through the MCP (Model Context Protocol) framework, enabling you to perform catalog operations, search Discogs collections, and interact with Discogs data via MCP clients such as Claude Desktop, LibreChat, and LM Studio. The server relies on a Discogs personal access token to authenticate requests to the Discogs API. You can run it locally (via Node.js/PNPM) or in a Docker container, and you can connect to it from MCP clients using the provided configuration formats (NPX, Local Node, or Docker).

To use the server with Claude Desktop, obtain your Discogs token and configure one of the client options shown in the README (NPX, Local Node, or Docker). The NPX option runs the server directly from the npm registry, supplying your token through environment variables. The server exposes a set of tools defined by the Discogs API, including catalog lookups, releases, artists, and collection editing operations, with explicit caveats noted in the documentation (for example, careful handling of edits and token security). When using the MCP Inspector, you can test endpoints and view responses as you build prompts and prompts to Claude or other clients.

Prerequisites:

• Node.js (tested with Node.js 20.x.x; Node 18.x.x should work)
• PNPM (recommended) or npm/yarn
• Optional: Docker if you want to run via Docker

Step-by-step installation:

1. Clone the repository (or use the published npm package): git clone https://github.com/cswkim/discogs-mcp-server.git cd discogs-mcp-server

2. Install dependencies using PNPM (preferred) or npm: pnpm install

   or npm install

3. Create a .env file in the project root based on .env.example and set required variables: DISCOGS_PERSONAL_ACCESS_TOKEN=<your_token> SERVER_HOST=0.0.0.0 # or 127.0.0.1 as needed

   Other optional environment variables can be configured as documented in .env.example

4. Run the server locally (development): pnpm run dev

   Access via your MCP client using the NPX method described in the README

5. Optional: Build and run a production build: pnpm run build pnpm run start

6. (If using Docker) Build and run: docker build -t discogs-mcp-server:latest . docker run --env-file .env discogs-mcp-server:latest

Note: If you plan to use HTTP streaming transport, ensure your .env contains the correct port mappings and that you run the container with the appropriate stream mode (as described in the Docker section of the README).

Helpful tips and common considerations:

• Your Discogs token should be kept secret. Do not share your token publicly and consider using environment variable management (e.g., .env files or secret managers) in deployment.
• The server defaults the Discogs API per_page to 5 for lighter responses, but you can adjust prompts to request more data; keep in mind some clients may struggle with larger responses.
• If you modify the server code, you may need to restart Claude or your MCP client to apply changes. Some tooling (like Claude) prompts for permission before running MCP tools; this is expected behavior.
• For Docker usage, you can run with an env-file and in streaming mode by mapping ports and using the stream argument, as shown in the README.
• The MCP Inspector is available to test endpoints locally at http://127.0.0.1:6274; use it to validate your server configuration before connecting clients.
• If you encounter type or endpoint issues, report them with reproducible steps and include the token scope and environment configuration used (without exposing tokens).