The useful boundary
The useful boundary is accountability without exposure. Heyrafiki publishes the interfaces and evidence that help institutions integrate and review behavior. Defensive controls stay private when disclosure would make impersonation, fraud or evasion easier.
A small set of controls stays private where publication itself would create the vulnerability:
- Safety and abuse controls. Published clinical scoring remains attached to each Instrument. Private abuse-detection signals, escalation operations and response controls stay restricted so they cannot be rehearsed against the system.
- Anti-Fraud logic. Insurance abuse drains Care from the People it was funded for, and a fraud model loses most of its power the day its rules are published.
- Verification pipelines. The sources and decision status are explainable to authorized reviewers. The exact anti-impersonation checks stay private so a false profile cannot prepare around them.
- Certain commercial integrations. Partner-specific implementation details stay private where legal, privacy or security obligations require it, and no further than necessary.
The test we apply
Every component faces the same question: does publishing it improve accountability and interoperability, or does it make abuse easier? Public value belongs in the open. Exploitable defensive detail does not.
Held accountable anyway
Private controls still produce reviewable evidence: who acted, under which authority, what decision was made and which version of the rule applied. Authorized auditors and regulators can examine that evidence without publishing the defensive logic itself.
The rule is consistent: publish what strengthens trust and integration; restrict what would expose People or help someone defeat a safeguard.
