Skip to main content
context is answer plus a coarse patient profile. It contextualizes the answer and surfaces conflicts between the recommendations and the patient — allergies, interactions with current treatments, pregnancy, age. Endpoint: POST /v1/context · MCP tool: context

​
The pseudo-context boundary

The patient profile is deliberately coarse. This is a pre-MDR boundary: the patient_context contract intentionally does not go more granular. It is a parameter of this capability, not a separate route.
FieldType
sexstring
ageint
historystring[]
active_problemsstring[]
allergiesstring[]
current_treatmentsstring[]

​
Request

{
  "query": "Quelle antibiothérapie pour cette infection urinaire ?",
  "patient_context": {
    "sex": "F",
    "age": 34,
    "active_problems": ["grossesse 24 SA"],
    "allergies": ["pénicilline"],
    "current_treatments": ["warfarine"]
  }
}

​
Response

Same shape as answer, plus a conflicts array: 
{
  "answer": "...",
  "sources": [{ "reco_id": 900, "reco_title": "...", "source_name": "HAS", "deep_link": "..." }],
  "conflicts": [
    {
      "description": "La molécule recommandée est contre-indiquée pendant la grossesse",
      "patient_factor": "grossesse 24 SA",
      "reco_id": 900
    }
  ]
}
Example worth knowing: a PSA-screening recommendation that conflicts with a patient’s active context is surfaced in conflicts rather than buried in prose.
Run it live in the API Reference tab.