The FHIR API for LATAM healthcare

Read and write clinical data over standards-compliant SMART on FHIR, modeled for the names, identifiers, and catalogs LATAM clinics actually use. First request in the browser, no key required.

Example console: a GET /fhir/R4/Patient request returns a 200 OK FHIR Bundle with a two-surname patient (María Elena Fernández) and a CURP identifier; the same search runs from the huli CLI as "huli fhir patient search". Sandbox credentials only — no production data.

Standards, by the book

SMART on FHIR R4 and R5 — the resources, scopes, and errors you already know.

Prototype before you commit

Your first request runs in the browser against a pre-seeded sandbox — no key needed.

LATAM modeling, already done

Two-surname names, CURP / RFC / NSS IDs, and Mexican address codes — built in.

Automation, agent-ready

Script every call with the huli CLI — JSON output, ready for CI and AI agents.

Playground

Run it live — no key required

Press Run to watch the search execute in your browser and step through the Bundle it returns.

Sandbox

Request

GET/fhir/R4/Patient?name=Fernández
Accept: application/fhir+jsonAuthorization: Bearer <sandbox key>

Press Run and watch this exact search execute in your browser — fabricated patients, no key, no sign-up — with a step-by-step walkthrough of the response.

Run
200 OKapplication/fhir+json
{
"resourceType": "Bundle",
"type": "searchset",
"total": 12,
"entry": [{
"resource": {
"resourceType": "Patient",
"name": [{ "family": "Fernández", "given": ["María", "Elena"] }],
"identifier": [{
"system": "https://www.gob.mx/curp",
"value": "FEME800614MDFRRR09"
}],
"gender": "female",
"birthDate": "1980-06-14"
}
}]
}

Recipes

What do you want to build?

Copy-paste paths from empty project to working integration.

All recipes

Platform

Everything an integration needs

A small, predictable surface: standard resources, scoped credentials, errors you can branch on.

Built for AI agents

Point your agent at these docs and it can hold the whole API in context. Every page has a plain-Markdown mirror — append .md to any URL — and the full corpus ships as a single file.

# everything an agent needs, three GETs

GET /llms.txt → index of every page

GET /llms-full.txt → the full corpus, one file

GET /v1/api/r4/patient.md → any page as plain Markdown

Your first 200 OK is one minute away

Run the request in the sandbox now — mint a key when you're ready to ship.