Skip to content

Guide Workflows

BioMCP ships three deterministic investigation workflows in its embedded agent guide. This page publishes those workflows directly in the docs so you can follow them from the CLI without installing the guide into an agent directory.

The command lines and workflow guardrails here mirror the embedded skills/SKILL.md. For the validation rubric that explains when a workflow run is complete and trustworthy, see Skill Validation.

Variant Pathogenicity Workflow

When to use: you have a specific variant and need a focused answer about pathogenicity and clinical evidence.

biomcp get variant "<id>" clinvar predictions population
biomcp get variant "<id>" civic cgi
biomcp variant trials "<id>"
biomcp variant articles "<id>"

Only add more commands if a needed claim is still unsupported.

Example:

biomcp get variant "BRAF V600E" clinvar predictions population
biomcp get variant "BRAF V600E" civic cgi
biomcp variant trials "BRAF V600E"
biomcp variant articles "BRAF V600E"

Drug Safety Workflow

When to use: you need a concise safety summary for a specific drug.

Quick safety summary:

biomcp get drug <name> label interactions approvals
biomcp drug adverse-events <name>

Filtered FDA adverse-event check:

biomcp get drug <name> label interactions approvals
biomcp search adverse-event --drug <name> --outcome death --limit 10

Do not write biomcp drug adverse-events <name> --outcome ....

Example:

biomcp get drug pembrolizumab label interactions approvals
biomcp drug adverse-events pembrolizumab

Broad Gene-Disease Workflow

When to use: you want broad but bounded context on a gene in a specific disease.

biomcp search all --gene <gene> --disease "<disease>" --counts-only
biomcp get gene <gene> pathways diseases protein druggability civic
biomcp search drug --target <gene> --indication "<disease>" --limit 10
biomcp search trial -c "<disease>" --mutation "<gene>" --status recruiting --limit 10
biomcp search article -g <gene> -d "<disease>" --sort citations --limit 10

Rules:

  • do not run search disease unless you need an ontology ID or phenotype sections
  • do not use free-text search drug when --target or --indication is enough
  • do not run both search drug <gene> and search drug --target <gene> in the same investigation
  • get variant only for simple substitutions or exact IDs copied from search results
  • do not get variant on exon-level free text like "Exon 19 Deletion"
  • for EGFR/NSCLC, cover exon 19 deletions and exon 20 insertions from disease, drug, trial, or article evidence unless an exact variant ID is surfaced
  • if you need a variant deep dive, choose at most two exemplar simple substitutions such as L858R and T790M
  • choose at most two exemplar variants for deep follow-up
  • choose at most three representative EGFR drugs for deep follow-up; do not fetch near-duplicates like both erlotinib and erlotinib hydrochloride unless the distinction matters
  • fetch only one or two key articles or trials unless the prompt explicitly asks for exhaustive evidence
  • stop once you can cover: gene role/pathway, actionable alterations, approved drugs, active trials, and resistance mechanisms

Example:

biomcp search all --gene EGFR --disease "non-small cell lung cancer" --counts-only
biomcp get gene EGFR pathways diseases protein druggability civic
biomcp search drug --target EGFR --indication "non-small cell lung cancer" --limit 10
biomcp search trial -c "non-small cell lung cancer" --mutation "EGFR" --status recruiting --limit 10
biomcp search article -g EGFR -d "non-small cell lung cancer" --sort citations --limit 10

Evidence Discipline

Across all workflows:

  • only claim facts the current command output supports
  • keep the IDs that make the evidence rerunnable, such as variant IDs, PMIDs, and NCT IDs
  • prefer source-tied phrasing such as ClinVar shows..., CIViC reports..., or the retrieved trials include...
  • if you need a mechanism, approval, or article detail, fetch the section or article that shows it
  • if an exact count or score is not clearly visible in the output, summarize qualitatively instead of guessing
  • avoid words like definitive or proves unless the retrieved evidence justifies that certainty
  • if one command already answers the question, stop searching

See Also