Built in Kenya, for the World.
Mental Healthcare should get better informed over time. Heyrafiki brings early Screening, finding care, ongoing Care, Benefits and Claims into one connected system, so a Person does not have to start from zero whenever care changes hands.
People decide what they share. Practitioners decide what it means. Organizations and Insurers act only within the permissions and responsibilities assigned to them.
Website · Docs · Lab · Open Source · Vision
| Foundation | What it makes possible |
|---|---|
| Continuity | The context a Person agrees to share can move from one point of Care to the next. |
| Intelligence | Repeated check-ins help a Practitioner see what changed and decide what needs attention. |
| Infrastructure | Care, Benefits, Claims and Payments can work together while each keeps its own responsibilities. |
| Repository | What it is |
|---|---|
| contract | The versioned public API contract. |
| proving-ground | Open tests that show what the system supports and where it must stop. |
| docs | Guides for the API, SDKs, Webhooks and MCP access. |
| rafiki-js | JavaScript and TypeScript client for the Heyrafiki API. |
| rafiki-py | Python client for the Heyrafiki API. |
| rafiki-go | Go client for the Heyrafiki API. |
| rafiki-net | .NET client for the Heyrafiki API. |
| rafiki-rs | Rust client for the Heyrafiki API. |
| hey | Command-line tools for safe Sandbox testing. |
Proving Ground publishes test suites that anyone can run against our public contracts. They cover consent and access, payment and Claim records, early-Screening provenance, and the cases in which the system must decline to act.
First Light keeps the questionnaire, version, language, timing, Consent and source with an early-Screening result. That gives a Practitioner the context to understand a result before relying on it.
Iris Affective Dynamics investigates whether repeated observations can help Practitioners notice deterioration, persistence and recovery sooner. Its comparators, failure criteria and evidence boundaries are published before clinical use.
- People decide what to share.
- Practitioners make clinical decisions.
- Consent determines access.
- Insurers make Claim decisions under their own rules.
- AI can help people make sense of information. It does not make clinical, Benefit or Claim decisions.
Our public work includes the API definition, SDKs, command-line tools, documentation and test suites. For agent integrations, MCP access starts in approved Sandbox projects with clear scopes and an audit trail.
Open source has a clear boundary. Production data, credentials, safety systems, and private regulatory or insurer integrations remain private. Read the open source boundary.
Start with CONTRIBUTING.md. Report vulnerabilities through SECURITY.md, never a public issue.