Your records,
AI-powered
Discogs MCP is an open-source Model Context Protocol server that connects Claude and other AI assistants to your Discogs record collection. Search by genre, decade, or just tell it you want something for a rainy Sunday. Self-hosted on Cloudflare Workers — you bring your own Discogs API key.
Heads up: discogs-mcp.com is the maintainer's private instance and is locked to a single Discogs account. The Discogs API rate limit is too tight to share, so you'll need to deploy your own copy (free on Cloudflare Workers, takes ~10 minutes).
Things you can ask
Something for a rainy Sunday afternoonMood-based recommendations from your actual collection NEW
What jazz records do I own?Searches your full Discogs library by genre, style, or artist
How many records from the 70s do I have?Collection statistics and breakdowns
Tell me about this pressing of Kind of BlueDetailed release info including label, year, and format
Setup
Deploy your own Discogs MCP server, then point your AI client at it.
1. Deploy your own instance
Clone the repo, add your Discogs API credentials, and deploy to Cloudflare Workers. The free tier is more than enough.
2. Connect your MCP client
Once deployed, you'll have a URL like https://discogs-mcp.<your-subdomain>.workers.dev/mcp. Point your MCP client at that URL — the same instructions work in Claude Desktop, Claude Code, Cursor, Windsurf, OpenCode, and any other MCP-compatible client.
The full configuration snippets are in the README.
Available Tools
Public (no auth)
- ping
- server_info
- auth_status
Search & discovery
- search_collection
- search_discogs
- get_release
- get_collection_stats
- get_recommendations
Collection management
- add_to_collection
- remove_from_collection
- move_release
- rate_release
Folders
- list_folders
- create_folder
- edit_folder
- delete_folder
Custom fields
- list_custom_fields
- edit_custom_field
Diagnostics
- get_cache_stats