Recipe: Drug safety profile

This recipe walks an agent from “ozempic” to a structured safety profile — RxNorm canonical concept, FDA adverse-event report counts, FDA approval history, active clinical trials — with pipeworx:// citations.

Tools used: resolve_entity, compare_entities (optional), fda_drug_events, fda_drug_approvals, ct_search, rxnorm_interactions.

Calls: 1 (compound) or 4 (parallel) depending on path.

Fast path: `compare_entities`

For single-drug or comparative profiles, compare_entities runs RxNorm + FDA + ClinicalTrials in parallel:

compare_entities({
  type: "drug",
  values: ["ozempic", "mounjaro"]
})

Returns:

{
  "type": "drug",
  "entities": [
    {
      "query": "ozempic",
      "resolved": true,
      "rxcui": "1991311",
      "rxcui_name": "...semaglutide... [Ozempic]",
      "facts": {
        "adverse_event_reports": 63000,
        "approval_count": 2,
        "active_trial_count": 104
      },
      "resources": {
        "rxnorm_concept":      "pipeworx://rxnorm/concept/1991311",
        "rxnorm_interactions": "pipeworx://rxnorm/concept/1991311/interactions"
      }
    },
    { "query": "mounjaro", ...same shape... }
  ],
  "deltas": {
    "adverse_event_reports": [{entity: "ozempic", value: 63000}, {entity: "mounjaro", value: 87308}],
    ...
  }
}

One call, three sources, parallel — replaces ~8 sequential calls.

Manual path (single drug)

If you want finer control:

Step 1: Resolve to RxCUI

resolve_entity({ type: "drug", value: "ozempic" })
// → { ids: { rxcui: "1991311", product_name: "..." },
//     resources: { rxnorm_concept, rxnorm_interactions } }

Step 2: Adverse events from FAERS

fda_drug_events({ query: "ozempic", limit: 10 })
// → { total: 63000, results: [{...full safety report...}, ...] }

The total field is the report count. The results array contains individual reports — useful to bucket by reaction type.

Step 3: FDA approvals

fda_drug_approvals({ query: "ozempic" })
// → { total: 2, results: [{approval_date, indication, ...}] }

Step 4: Active clinical trials

ct_search({ query: "ozempic", limit: 10 })
// → { total_count: 104, studies: [{nct_id, brief_title, phase, status, ...}] }

For each interesting trial, embed pipeworx://clinicaltrials/study/{nct_id} in your output.

Step 5: Drug interactions

rxnorm_interactions({ rxcui: "1991311" })
// → list of clinically significant interactions

Or via resource URI: resources/read({ uri: "pipeworx://rxnorm/concept/1991311/interactions" }).

Citation pattern

Embed in your output:

Ozempic (RxNorm 1991311) has 63,000 adverse-event reports in FAERS, 2 FDA approvals (2017 for type 2 diabetes, 2024 for cardiovascular risk reduction), and 104 active or recruiting clinical trials at ClinicalTrials.gov. Notable interaction class: insulin secretagogues — see interactions.

The pipeworx:// URIs are resolvable by any MCP client supporting Resources.

Use the prompt

For the full playbook with prose framing:

prompts/get({
  name: "drug_safety_profile",
  arguments: { drug_name: "ozempic" }
})

Returns a substituted prompt that orchestrates the full sequence including a structured executive summary.

Caveats

Branded queries return SBD only. Searching “ozempic” returns Semantic Branded Drug concepts. To get the ingredient (semaglutide) RxCUI, run rxnorm_search({name: "semaglutide"}) separately.
Adverse-event totals are reports, not unique patients. A single hospitalization can generate multiple reports.
Trial counts include all statuses. Filter studies[].status to “Recruiting” / “Active, not recruiting” for active research only.