{
  "markdown": "# Noodle Biomedical Literature Discovery MCP\n\n[![DOI](https://zenodo.org/badge/DOI/10.5281/zenodo.22166486.svg)](https://doi.org/10.5281/zenodo.22166486)\n[![AllMCPs Verified](https://allmcps.com/api/badge/noodle-biomedical-literature-discovery-mcp)](https://allmcps.com/mcp/noodle-biomedical-literature-discovery-mcp?verify=930d811d-1fd8-4117-8280-1c6eac1a2ca3)\n[![OpenSSF Scorecard](https://api.scorecard.dev/projects/github.com/helena-bioinformatics/noodle-mcp/badge)](https://scorecard.dev/viewer/?uri=github.com/helena-bioinformatics/noodle-mcp)\n[![RRID](https://img.shields.io/badge/RRID-SCR__028920-005A9C)](https://n2t.net/RRID:SCR_028920)\n\nThe official public, read-only Model Context Protocol adapter for biomedical\nliterature discovery from [Helena Bioinformatics](https://www.helena.bio).\nAgents can select it from a user task even when the user does not know the\nNoodle brand.\n\nPublic endpoint: `https://api.helena.bio/noodle/v1/mcp`\n\nOfficial Registry identity: `io.github.helena-bioinformatics/noodle`\n\nNo account, API key, patient data, or private content is required or accepted.\n\n## What agents can do\n\n- search a public PubMed-derived biomedical corpus by natural language, PMID,\n  DOI, or PMCID;\n- retrieve source-linked publication records by PMID or Noodle work ID;\n- traverse bounded citation and semantic neighborhoods from a publication;\n- continue graph exploration through returned work identifiers while\n  preserving edge types and graph provenance;\n- inspect corpus size, sources, freshness, coverage, and active graph metadata.\n\nThe seven published tools are `search_biomedical_literature`,\n`get_publication_details`, `get_work_details`,\n`get_publication_neighborhood`, `get_work_neighborhood`,\n`get_corpus_summary`, and the separate explicit opt-in `support_helena`\ninformation action.\n\n## Connect\n\nAny MCP client that supports remote Streamable HTTP can use the endpoint. Exact\nrecipes for ChatGPT, Claude, Codex, VS Code, Cursor, Windsurf, Gemini CLI,\nGrok, Perplexity, Microsoft Copilot Studio, Biomni, and Biorouter live under\n`registry/platforms` and `integrations`.\n\nReady-to-use ecosystem packages are included for:\n\n- [Dify](integrations/dify/README.md), including a reproducible `.difypkg`;\n- [n8n](integrations/n8n/README.md), using the exact stateless MCP JSON-RPC\n  contract supported by the hosted service;\n- [Galaxy](integrations/galaxy/README.md), with a Planemo-linted ToolShed\n  wrapper; and\n- [KNIME Analytics Platform](integrations/knime/README.md), with a table-to-MCP\n  Python Script node and prepared Hub listing; and\n- [Google Colab and Kaggle notebooks](integrations/notebooks/README.md), plus a\n  [Cytoscape GraphML workflow](integrations/cytoscape/README.md); and\n- the companion [Galaxy Training Network tutorial](integrations/galaxy-training-network/README.md)\n  for a Folklore-to-Noodle literature workflow.\n\nThe companion Agent Skill is in\n`skills/noodle-biomedical-literature-discovery`. It enables implicit,\ntask-first selection for requests such as:\n\n- “Find source-linked papers about BRCA1 homologous recombination.”\n- “What publication is PMID 35008774?”\n- “Show papers related to this article through citations and semantic\n  similarity.”\n- “Walk two bounded hops from this work ID and preserve the edge types.”\n\nBuild the deterministic skill archive with:\n\n```bash\npython3 ops/package_agent_skill.py\n```\n\n## Graph boundary\n\nStart from a resolved PMID or work ID and request one bounded neighborhood at a\ntime. Report edges exactly as returned, keep a visited-ID set, and stop at a\nmissing neighborhood. Search rank, citation proximity, semantic similarity,\nco-mention, and graph distance are discovery signals. They do not establish\ncausality, scientific validity, diagnosis, or treatment.\n\n## Development\n\nPython 3.12 is required.\n\n```bash\npython -m venv .venv\n. .venv/bin/activate\npython -m pip install -r requirements-dev.lock\npython -m pip install --no-deps -e .\npytest\nruff check .\nruff format --check .\n```\n\nRun the brand-blind contract audit with:\n\n```bash\npython benchmarks/agent-discovery/audit_skill.py\n```\n\nThe benchmark contains 60 prompts that omit `Noodle`, `Helena`, and `MCP`.\nIt covers all six scientific routes plus negative and safety controls.\n\n## Agent Plugin and Kiro Power\n\nThis repository is also a portable Agent Plugin and Kiro Power. `plugin.json`\nprovides brand-blind activation keywords, the existing Agent Skill supplies the\nscientific routing and safety boundary, and `mcp.json` connects directly to the\ncanonical hosted Streamable HTTP endpoint. The Power does not proxy, repackage,\nor reimplement Noodle.\n\nPrivacy policy: https://noodle.helena.bio/privacy\n\n## Cite Noodle\n\nThe persistent Research Resource Identifier is\n[`RRID:SCR_028920`](https://n2t.net/RRID:SCR_028920). Cite the resource in a\nmethods section as **Noodle (RRID:SCR_028920)**. Use the\n[version DOI](https://doi.org/10.5281/zenodo.22166486) when a version-specific\nsoftware citation is also needed. The RRID identifies the resource across\npublications, while the DOI identifies the archived 0.2.0 release.\n\nSupport: https://noodle.helena.bio/contact or `contact@helena.bio`\n\n## Public resources\n\n- Hands-on tutorial: https://github.com/helena-bioinformatics/noodle-mcp/blob/main/docs/tutorials/biomedical-literature-discovery-and-graph-traversal.md\n- Connector and agent-selection guide: https://noodle.helena.bio/mcp\n- Client integrations: https://noodle.helena.bio/integrations\n- Server Card: https://noodle.helena.bio/.well-known/mcp/server-card.json\n- Official Registry: https://registry.modelcontextprotocol.io/v0/servers?search=io.github.helena-bioinformatics%2Fnoodle\n- Citable release: https://doi.org/10.5281/zenodo.22166486\n- Software Heritage archive request: https://archive.softwareheritage.org/api/1/origin/save/2457442/\n- Software Heritage snapshot: https://archive.softwareheritage.org/swh:1:snp:09b8fb7c64de15487e873b4f77e3e4b57abc02fb/\n- Methodology: https://noodle.helena.bio/methodology\n\n## License and security\n\nApache License 2.0. Report vulnerabilities privately as described in\n`SECURITY.md`. Do not submit patient, private case, clinical-record, credential,\nor private uploaded content to the public service or issue tracker.\n",
  "bytes": 6220,
  "sha": "817ab3eac3bafd276aac21db1368bfa0c4bdb237c1cf9aeea72bdfa077fd8337",
  "repo_slug": "helena-bioinformatics/noodle-mcp",
  "fonte": "repo",
  "truncated": false,
  "api": "https://agentalog.com/api/listings/mcp_io_github_helena_bioinformatics_noodle_e75c7245/readme"
}