Health intelligence API

Health analysis API for agents.

Send a raw VCF, a lab panel, or a wearable export. Get back normalized observations, clear interpretations, and a dashboard spec — over REST or MCP, in minutes. No bioinformatics team required.

Get an API key →Self-host for free
MIT-licensed core· EU-hosted· MCP-native· OpenAPI 3.1· 125 tests passing
POST /v1/analyses
# one call, three modalities
curl -X POST https://api.foreverbetter.xyz/v1/analyses \
  -H "Authorization: Bearer $FB_TOKEN" \
  -d '{ "source_ids": ["src_vcf_8f2a", "src_whoop_0431"] }'

200 OK
{
  "observations": 142,
  "interpretations": [
    { "id": "apob",    "level": "attention" },
    { "id": "vo2max",  "level": "good" },
    { "id": "cyp2c19", "level": "pgx" }
  ],
  "dashboard_spec": "ds_19c4f0"
}

Works with the data your users already have

VCF / WGS23andMeAncestryDNADante LabsNebulaWHOOPOuraApple HealthGarminFitbitWithingsBlood panel CSVQuestSYNLABClinVarCPICPGS Catalog
The hard part

Health data is brutal to build with.

Genetics, labs, and wearables each arrive in their own messy format. Turning them into something you can trust is months of bioinformatics and compliance work — before you ship a single feature.

You'd build · Genetics

Parse a VCF, then map ClinVar findings, CPIC pharmacogenomics, and polygenic scores yourself.

You'd build · Wearables

Reconcile WHOOP against Oura against Apple Health — every vendor exports a different shape.

You'd build · Biomarkers

Score every lab marker against reference ranges, track trends, and keep the whole thing compliant.

The fix

Genetics, wearables, and biomarkers — composed.

Send one modality or all three. The same composer returns a single structured picture, so the genome quietly tailors what the labs and wearables already show.

Genetics

Whole genomes to insights

Send a VCF, WGS export, or 23andMe / AncestryDNA file. Get ClinVar findings, CPIC pharmacogenomics, polygenic scores, and aging-hallmark interpretation back as structured data.

  • VCF / VCF.GZ
  • WGS
  • SNP array
  • ClinVar
  • CPIC
  • PRS
Wearables

Recovery, sleep, strain

Connect WHOOP or Oura, or upload Apple Health / Garmin exports. Readings are normalized into HRV, resting HR, sleep stages, recovery, and SpO₂ observations — ready to score.

  • WHOOP
  • Oura
  • Apple Health
  • Garmin
  • OAuth sync
Biomarkers

Labs that mean something

Upload blood panels as CSV, JSON, or a lab export. Each marker is scored against reference ranges with trends and plain-language flags — not just numbers on a page.

  • CSV / JSON
  • Lab export
  • Reference ranges
  • Trends
Two ways to run it

Call the API, or own the whole pipeline.

Same primitives either way: raw source references, normalized observations, and derived interpretations.

Hosted API · fastest

Hosted API

A managed endpoint with auth, storage, and workers handled for you. Start sending data in minutes.

  • OIDC / JWT auth with per-endpoint and per-org scopes
  • Encrypted Supabase storage for raw payloads
  • Queue workers for full-size WGS jobs
  • MCP + REST, idempotency keys, and audit logs
Get an API key →Read the docs
Open source · local-first

analyze-health skill

The same engine as an agent skill that runs on your machine. Raw files never leave the device.

$npx skills add JamesLawton/agentic-health-analysis
  • Runs locally in your agent runtime — no account needed
  • Self-contained HTML dashboard output
  • MIT-licensed, auditable end to end
Star on GitHubRead SKILL.md
Prebuilt flows

The endpoints you'd have built anyway.

Every route also works behind a /v1/ prefix and is published in an OpenAPI 3.1 contract.

POST/imports/fileUpload genetic, biomarker, wearable, or behavioral data.
POST/analysesRun a multi-modal analysis across uploaded source IDs.
POST/queryQuery normalized observations and derived interpretations.
GET/dashboard-specs/:idPull a dashboard-ready JSON spec for any analysis.
POST/connections/wearables/syncPull connected WHOOP / Oura data on demand or on a schedule.
GET/labs/searchLocate Quest Diagnostics and SYNLAB draw sites.
POST/mcpThe same primitives, exposed as MCP tools for agents.
How it works

From raw file to structured insight.

  1. 01

    Get a key

    Request access and issue a scoped, short-lived JWT. Limit each token to the exact endpoints and organization an agent should touch.

  2. 02

    Send data

    Upload a file or connect a wearable — over plain REST or as an MCP tool call. Full WGS runs queue automatically to a worker.

  3. 03

    Get structure back

    Receive normalized observations and interpretations with severity levels and provenance — never raw, unexplained numbers.

  4. 04

    Render or reason

    Pull a dashboard spec to render a report, or feed the interpretations straight into the agent you're building.

Agent-native

Built for agents, not just apps.

Discover the contract at /.well-known/health-agent.json, list the tools a token is allowed to use, then call them. The tool list is filtered to each token's scopes — so an agent only ever sees what the user enabled.

Read the MCP guideView the OpenAPI spec
POST /mcp · tools/call
{
  "method": "tools/call",
  "params": {
    "name": "query_health_context",
    "arguments": {
      "user_id": "u_204",
      "query": "ApoB"
    }
  }
}
Who builds on it

For the people putting health data to work.

Health & wellness SaaS

Ship health features, skip the bioinformatics team

Add genetics, labs, and wearables to your product through one API. Per-organization scoping and audit logs are built in.

Longevity clinics & orgs

Standardize intake, own your infra

Normalize multi-modal member data and generate dashboards on your own deployment — EU-hosted, with deletion and export workflows.

Agent builders

Give your agent real health context

Discover the contract, list the tools a token allows, and call them. Scoped, auditable access with no glue code.

Questions

Answers before you ask.

Is this medical advice?

No. foreverbetter is a wellness and healthspan layer, not a diagnosis, medical device, or treatment engine. High-stakes findings should be confirmed with a clinician or genetic counselor.

Do I have to use the cloud?

No. The analyze-health skill is MIT-licensed and runs locally — raw files never leave the machine. The hosted API is for teams who would rather not run the infra.

Where does the data live?

The hosted API is EU-hosted with encrypted storage, per-organization scoping, and deletion and export workflows. Self-hosted, nothing leaves your environment.

What formats do you accept?

VCF / WGS and 23andMe / AncestryDNA for genetics, CSV / JSON / lab exports for biomarkers, and WHOOP / Oura / Apple Health / Garmin for wearables.

How do agents authenticate?

Short-lived JWTs scoped to specific endpoints and a single organization. An agent's tools/list returns only the tools its token is allowed to call.

What does it cost?

The open-source skill is free to self-host. The hosted API is usage-based — get a key to start, no credit card required.

Start building with health data.

One API for genetics, wearables, and biomarkers. Or the open-source skill, free. Send your first file today.

Get an API key →Read the docs

Open source and free to self-host · no credit card to try the API

Looking for the membership for individuals? Visit members.foreverbetter.xyz →