For AI agents
Search papers and traverse related biomedical literature.
Noodle Biomedical Literature Discovery MCP is Helena Bioinformatics' free public read-only interface for literature search, publication inspection, and bounded citation or semantic graph exploration.
Connect
Use Streamable HTTP at https://api.helena.bio/noodle/v1/mcp. Authentication is not required. The stable Official MCP Registry identity is io.github.helena-bioinformatics/noodle.
Open the Official MCP Registry record or read the complete Server Card.
Tasks agents can discover
Use the server when a user asks to find biomedical papers, resolve a PMID, DOI or PMCID, inspect a publication or corpus work, find related research, walk a citation or semantic literature neighborhood, or check corpus coverage and freshness. The user does not need to know the Noodle name.
Graph traversal
Start from a returned PMID or work identifier and request one bounded neighborhood at a time. Preserve every returned edge type and graph version. Continue with the selected neighbor as the next anchor, keep a visited set, and stop at a missing neighborhood. Citation proximity and semantic similarity are discovery signals, not proof of causality.
Safety and provenance
Send only public non-sensitive questions and identifiers. Never send patient records, private case data, credentials, or private files. Preserve source links and verify important conclusions in the primary publications. Noodle does not diagnose patients or recommend treatment.
Install task-first selection
The public companion Agent Skill is named noodle-biomedical-literature-discovery. It declares implicit invocation for paper search, publication identifiers, related-paper discovery, citation graphs, semantic neighborhoods, and corpus questions.