One rail for Care, Cover and Payments. Built in Nairobi, for Africa and the next billion People.
Website · Docs · Changelog · Vision · Open source
Heyrafiki connects People, Practitioners, Insurers and Organizations across Care, Benefits and Payments. Developers build against a versioned REST contract, typed SDKs, the hey command line and public documentation.
| Repository | Purpose | Licence |
|---|---|---|
docs |
Developer documentation for the API, SDKs, Webhooks and MCP. Read the Docs. | CC BY 4.0 |
contract |
OpenAPI 3.1 contract and machine-readable Assurance Graph. | Apache 2.0 |
proving-ground |
Executable contract, financial, bitemporal and adversarial conformance evidence. | Apache 2.0 |
rafiki-js |
JavaScript and TypeScript SDK for the Heyrafiki API. | Apache 2.0 |
rafiki-py |
Python SDK for the Heyrafiki API. | Apache 2.0 |
rafiki-go |
Go SDK for the Heyrafiki API. | Apache 2.0 |
rafiki-net |
.NET SDK for the Heyrafiki API. | Apache 2.0 |
rafiki-rs |
Rust SDK for the Heyrafiki API. | Apache 2.0 |
hey |
Command line for Sandbox diagnostics and governed API reads. | Apache 2.0 |
The documentation, API contract, assurance benchmark and SDK source are public. Organizations with Sandbox access can create a key in the Developer Platform. Request access.
Integration teams can review the contract and control model before requesting access:
| Review surface | What it proves |
|---|---|
| Insurance integration guide | Coverage, eligibility, pre-authorization, Claims and remittance workflow ownership |
| Financial controls | Integer money, Claim line identities, decision derivation and settlement separation |
| Claim valuation timeline | Reproducible as-of Claim history across business time, knowledge time and valuation time |
| Assurance Graph | Every public operation linked to an accountable capability, control owner, authority source and executable evidence |
| Acceptance testing | Synthetic pilot sequence, negative tests and production approval evidence |
| OpenAPI contract | Versioned schemas, scopes, errors, idempotency and accountable capability ownership |
| Proving Ground | Reproducible conformance suites and pinned evidence artifacts |
| First Light Evidence Pack | Consent-authorized early-identification routing, tenant isolation, blind dual review, longitudinal knowledge time and adverse-boundary evidence |
| Security policy | Private vulnerability reporting and health-data handling rules |
The payer remains authoritative for membership, Benefit design, adjudication policy, premium, reserving and regulatory returns. Heyrafiki preserves the operational evidence connecting payer decisions to delivered Care, Claims and settlement.
The Open Mental Health Insurance Assurance Benchmark maps all 31 public API operations to 10 accountable capabilities and 9 controls. Its deterministic runner validates the contract, financial identities, a bitemporal Claim valuation timeline and graph integrity. Five adversarial mutations test duplicate events, reordered knowledge, future knowledge, unbalanced adjudication and settlement above payer liability.
git clone https://github.com/heyrafiki/proving-ground.git
cd proving-ground
npm ci
npm testThe open benchmark is the shared technical baseline for governed institutional pilots. Teams extend it with their operating assumptions, approved evidence and accountable reviewers while preserving the same reproducible contract. Call the Claim valuation operation, read the research article or inspect the machine-readable Assurance Graph.
First Light carries Screening signals into an accountable Practitioner review boundary. Its public evidence pack reproduces 48 versioned fixtures across self, caregiver and Practitioner responder paths. The suite verifies Consent before read or action, tenant isolation, minimum-necessary model context, blind dual review, effective and recorded time, complete audit evidence and 24 adverse boundaries.
Every artifact is pinned in a reviewed SHA-256 manifest. The Evidence Boundary keeps fixed-cohort conformance measurement precise while instrument- and population-specific clinical evaluation carries its own reviewed protocol.
We publish contracts, SDKs and documentation where independent inspection improves interoperability and trust. Each public artifact has its own licence.
Safety thresholds, anti-fraud controls, private regulator and payer adapters, production credentials and production data stay private. Read the open source thesis.
Issues and pull requests are welcome in public repositories. Start with CONTRIBUTING.md.
Report security issues through SECURITY.md, never a public issue.