| name | biomed-research |
| description | Use when answering biomedical research questions that need source-backed evidence from local MCP servers, including gene, disease, drug, variant, phenotype, study, or clinical-trial questions |
Biomed Research
Use this workflow for biomedical research answers over the five local MCP servers.
Server Routing
| Need | Start With |
|---|
| target-disease evidence, drugs, variants, studies, trials | opentargets |
| phenotypes, HPO, rare disease, model organisms | monarch |
| gene annotation, IDs, expression, GO, orthologs | mygene |
| chemical/drug identifiers, mechanisms, structures | mychem |
| disease annotation, OMIM, Orphanet, MONDO, HPO | mydisease |
Resolve names to stable IDs first. Use the narrowest server that can answer the question. Fan out only when the question spans multiple entity types.
Evidence Rules
- Prefer curated MCP tools over raw schema queries.
- Use source-backed observations before synthesis.
- Keep result payloads small by requesting specific fields when tools support it.
- Reconcile conflicts explicitly instead of averaging them away.
- Stop after 1-2 follow-up rounds if gaps remain; report the limitation.
Output Contract
Return:
answer: concise biomedical synthesis.confidence: high, medium, or low.citations: up to 8 {observation_id, tool, note} entries.limitations: missing evidence, conflicts, source caveats, and safety notes.safety_note: included for personalized medical advice requests.
Use low confidence when no MCP evidence was used or sources conflict.
Safety Boundary
This is research and educational support, not clinical decision support. Do not diagnose, prescribe, dose, or recommend patient-specific treatment.
High-risk phrasing includes diagnose, treatment, prescribe, dose, dosage, should I take, should I stop, medical advice, my symptoms, and for me personally.
For those requests, provide only general biomedical context, decline personalized medical advice, and include the safety note in limitations.
