The README is aimed at researchers rather than engineers, and that is the useful signal: one person deploys the Worker, everyone else pastes a URL. Note that the URL in the documented config is the author's own example — replace it with your deployment's address, or the config will point somewhere you do not control.
An MCP server in front of the Drug Gene Interaction Database. It talks to DGIdb's public GraphQL API and hands the results to an assistant, so questions about how drugs interact with genes can be asked in a sentence rather than a query.
- Drug–gene interaction lookups from plain questions — the README's own examples are "What are the interactions for the drug Imatinib?", "Find genes that interact with Dovitinib." and "Show me drug attributes for Trametinib."
- Data straight from DGIdb, which aggregates drug–gene interaction evidence from many sources
- A deployment shape aimed at teams: run it on Cloudflare Workers once and give researchers a URL
- A client entry that needs no local toolchain — `npx mcp-remote` pointed at the deployed `/mcp` URL
For a researcher: the URL of a deployed instance, added to the Claude Desktop config through `mcp-remote`, then a restart. For whoever deploys it: a Cloudflare Workers account, with the server code in `src/index.ts` defining the DGIdb connection and the tools it offers. The upstream is the public DGIdb GraphQL API at `https://dgidb.org/api/graphql`, so no DGIdb account is involved. The same deployed URL also works from clients such as the Cloudflare AI Playground.
One command — npx mcp-remote https://dgidb-mcp-server.quentincody.workers.dev/mcp