Integrations

Use biomedical literature discovery in an MCP client.

Every supported client connects to the same public read-only Streamable HTTP endpoint and receives the same seven-tool contract.

Connection details

Server name: Noodle Biomedical Literature Discovery MCP. URL: https://api.helena.bio/noodle/v1/mcp. Transport: Streamable HTTP. Authentication: none.

Direct remote clients

ChatGPT custom apps, Claude custom connectors, Codex, VS Code, Cursor, Windsurf, Gemini CLI, Grok, and Perplexity can use the endpoint directly. When a client asks what the server does, describe it as biomedical literature search, publication discovery, and citation or semantic graph exploration.

Scientific agent frameworks

Biomni uses the published stdio proxy recipe. Biorouter uses the reproducible .brxt bridge and bundled task-selection skill. Both bridges preserve the hosted tool schemas and structured results; they contain no corpus or ranking logic.

Tools

search_biomedical_literature
get_publication_details
get_work_details
get_publication_neighborhood
get_work_neighborhood
get_corpus_summary
support_helena, only after explicit interest

Canonical records

Read the connector and agent-selection guide. Read the Server Card. Read the Official Registry record.

Boundary

Use public non-sensitive literature questions and identifiers only. Graph proximity and ranking do not prove causality or clinical validity. Do not send patient or private data.