API principle
CortexTPS endpoints treat every request as untrusted until validated. Actions are idempotent, scoped to the caller, and produce receipts.
Build against movement, receipts, and verified state — not vibes.
CortexTPS endpoints treat every request as untrusted until validated. Actions are idempotent, scoped to the caller, and produce receipts.
Route locks, proof saves, prompt shield blocks, and memory writes generate receipt hashes tied to a policy version. Receipts let callers verify a server action happened without exposing internals.
Integrations request capabilities, not raw authority. The server validates whether a capability is allowed in the current state of the room.
POST /api/public/privacy-request · POST /api/public/security-report · GET /api/public/trust-health
POST /api/public/privacy-requestPOST /api/public/security-reportGET /api/public/trust-healthAll endpoints are JSON. Successful writes return a tracking ID and a short receipt prefix. The trust-health endpoint exposes only boolean-style status flags — never secret values.