AI Tools / Healthcare

Public health data for your agent.

One MCP endpoint for deterministic retrieval across public biomedical sources, with source IDs, excerpts, limits, and citations kept in the response.

Access

Public beta. No model credentials are sent to OpenNash. Your MCP client supplies its own inference.

Connect

Add a remote HTTP MCP server in your MCP-compatible client and use this URL:

Endpointhttps://api.opennash.com/healthcare/mcp
{
  "mcpServers": {
    "opennash-health": {
      "type": "http",
      "url": "https://api.opennash.com/healthcare/mcp"
    }
  }
}

Live endpoint. Remote HTTP clients must support the MCP streamable HTTP transport and accept text/event-stream.

What it covers

  • Biomedical literature and open full text: PubMed, PMC, Europe PMC, Crossref, OpenAlex, and recent preprints.
  • Clinical trial designs, sites, status, results, versions, and eligibility text from ClinicalTrials.gov.
  • Medication identity, current labels, enforcement records, and spontaneous adverse-event signals.
  • Exact public UDI/device lookup, consumer guidance, terminology, public research awards, variants, genes, and proteins.

Useful requests

  • “Find recruiting phase 2 trials for this disease and return sites, eligibility text, NCT IDs, and source links.”
  • “Compare the current DailyMed label with FDA enforcement records for this medication. Separate label facts from safety signals.”
  • “Resolve these PMID, DOI, NCT, RxCUI, and UDI identifiers without fuzzy substitution.”
  • “Build an evidence table for this biomarker using literature, trials, ClinVar, NCBI Gene, and UniProt.”

Safety boundaries

Use public-data questions only. Do not enter names, dates of birth, medical record numbers, or other PHI. This tool does not diagnose, recommend treatment changes, determine trial eligibility, interpret medical images, or replace a clinician.

Search text is sent to the selected public source. Durable run records redact user query text and upstream-echoed raw fields, but that is not permission to submit sensitive data.

Verification model

Results carry stable source identifiers and known limitations. The service retrieves bounded evidence packets; it does not silently treat model prose as evidence. Registry-only sources are reported as unavailable rather than fabricated.

Building a regulated healthcare agent or evidence workflow? [email protected] →