Security and trust

Exceptional security must be measurable.

Implemented locally

Required before the health app can launch

AI boundary

The design does not give a language model database credentials, vault decryption authority or autonomous health-action authority. Numeric or comparative answers must resolve to evidence references or abstain.

Reporting a security concern

A public vulnerability-disclosure channel has not yet been activated. Do not send sensitive findings or health data through ordinary public channels. That channel is a launch prerequisite and will be published here when independently verified.