# Read and write patients and observations in an organisation's own FHIR server

> Read and write patients and observations in an organisation's own FHIR server: 0 companies fully do it, 4 with a limit, 11 carry the tag. Every listing in Health records & wellness data with the reason and a source.

- Canonical: https://www.anchorterminal.com/niches/health-data.fhir-store/
- Markdown: https://www.anchorterminal.com/niches/health-data.fhir-store/index.md (~1,250 tokens)
- Slim: https://www.anchorterminal.com/niches/health-data.fhir-store/index.min.md (~280 tokens, same facts, less prose, for token-sensitive contexts)
- JSON: https://www.anchorterminal.com/niches/health-data.fhir-store/index.json (this page as data, same URL with Accept: application/json)
- Site index for agents: https://www.anchorterminal.com/llms.txt (the key docs and every listing in one file, at most 16 MB, about 4 million tokens: https://www.anchorterminal.com/llms-full.txt)
- API: https://www.anchorterminal.com/api/v1/index.json
- MCP: https://www.anchorterminal.com/mcp (the whole catalogue as tools, no key)
- DNS: dig +short TXT <slug>.grade.anchorterminal.com (the grade as a DNS TXT record, for a sandbox with no HTTP where names still resolve; also top., best.<category>. and <a>.vs.<b>. under the zone, and the same under grade.letme.dev)
- Licence: CC BY 4.0, credit Anchor Terminal and link this page (https://www.anchorterminal.com/terms/#data)
- Updated: 2026-11-15

No company's docs show a full fit yet. The listings below do the job with a limit, or not at all.

| Companies, full fit | With a limit | Carry the tag | Share of the category's work | First-hand accounts |
| --- | --- | --- | --- | --- |
| 0 | 4 | 11 | 22 | 8 |

## What counts as done

A search returns only the requested resources and fields with working paging links, a conditional create on an identifier does not duplicate the patient, a write the policy forbids is refused with an OperationOutcome, and the written observation reads back unchanged.

Measured, for our own tests: a patient summary read within 20,000 tokens.

How agents ask for it: find this patient's last three blood pressure readings; record this observation; create the patient if they do not exist; what does this patient's chart say

## Does the job, with a limit

It does the job, but something in the docs or a user's account will get in an agent's way. The limit is named in each row.

- **1upHealth (1upHealth, Inc.)**: The 1up FHIR Server validates posted resources against base R4 (400 on invalid), writes Provenance on every create and update, and publishes /r4/metadata. It is a payer platform under an order form; conditional create, _elements and access-policy docs were not found in the pages read.
- **Aidbox (Health Samurai, Inc.)**: Search takes _elements, _count and _page with first, next and last links, the MCP has conditional update and create with headers plus validate, and SMART scopes and AccessPolicy deny with 403. The MCP module is alpha, and scopes are enforced only when the token's atv claim is 2.
- **Medplum (Orangebot, Inc.)**: Search pages with _count up to 1,000 and _cursor next links, access policies grant read-only per resource, and fhir-request does full CRUD with conditional headers. But fhir-request returns the raw bundle, which overflowed context for a real patient, and search and fetch are stubs (acc-15, issue 9616 open).
- **Zus Health (Zus Health, Inc.)**: The FHIR R4 API supports conditional create, update and delete by identifier, Zus-If-Match on updates, transaction Bundles and summarising lenses, and writes go back for the builder's own patients. No error body format or read-only scope enforcement was found in the pages read.

- not this job: Health Gorilla (Health Gorilla, Inc.): A network retrieval and lab API with writeback for supported resources, not an organisation's own FHIR store.
- not this job: Medblocks (Medblocks Inc.): Medblocks stores pulled records and can export them to a destination FHIR R4 server, but it is not an organisation's own read-write FHIR store.
- not this job: Metriport Medical API (Metriport Inc.): Consolidated data can be read and written as a Bundle, but every API key has full access including destructive operations, so a policy-forbidden write cannot be refused.
- not this job: Particle Health (Particle Health, Inc.): Returns retrieved data as FHIR datasets; it is not an organisation's own read-write FHIR store.
- not this job: Redox (Redox, Inc.): Redox translates and routes data between systems and destinations; it is not the organisation's own FHIR store.
- not this job: ROOK (Rookeries Development Corp.): Not a FHIR server; data arrives in ROOK's own sleep, physical and body schema.
- not this job: Terra API (Terra Enabling Developers Inc.): Not a FHIR server; data uses Terra's own data models.
- not this job: Validic Inform (Validic, Inc.): Not a FHIR server; records use Validic's own summary, sleep, workout and measurement resources.

## What stops agents finishing

- MCP returns raw bundles that overflow the context
- MCP tool can delete or overwrite anything the token allows
- self-hosted OAuth for a CLI client not documented

## Around it

Before: health-data.network-records, health-data.patient-records. After: scheduling.book, messaging.sms.

## Where this comes from

Weight 22 of 100 in Health records & wellness data: 8 of 36 use-case citations across 34 accounts, scaled to 100

As data: https://www.anchorterminal.com/api/v1/use-cases/health-data.fhir-store.json, or the MCP tool get_use_case with id health-data.fhir-store. Draft, read 2026-11-15 from public docs and first-hand accounts, not yet reviewed by a person.

---

Query this catalogue directly over MCP: https://www.anchorterminal.com/mcp (Streamable HTTP, no key). How: https://www.anchorterminal.com/agents/#mcp
