Reference

Glossary

One word, one meaning, everywhere: the API, the console, the demo, and these pages all read this file.

Terms

Term	Meaning
Brainprint	A private, tenant-scoped, continuously learned behavioral model that contributes experimental evidence about operator continuity. It is not a portable identity or a map of the brain.
Subject	A tenant-scoped pseudonymous person record that owns a Brainprint.
Evidence	Derived behavioral features and integrity metadata submitted by the SDK. Never raw content.
Assurance Profile	Separate typed claims about continuity, change, automation, approval, and unsupported capabilities, each with status, coverage, freshness, evidence references, reason codes, and limitations.
Protected action	A customer-registered consequential operation with a stable key, risk class, resource type, context schema, and policy.
Policy	The immutable, versioned requirements a protected action places on assurance evidence.
Decision	Exactly one of allow, challenge, hold, or deny, with stable reason codes. Insufficient evidence is an assurance status translated by policy.
Challenge	An action-bound passkey or labeled sandbox confirmation that re-runs the original authorization request without erasing a firing detector.
Proof of Agency	A short-lived, asymmetrically signed, audience-bound, action-bound artifact a relying party can verify without Brainprint data.
Evaluation	A time-boxed shadow-mode study with written success criteria.
Outcome	What the customer actually did after a decision.

What people see

Decisions and modes are shown to end users in plain words. Your application may use its own wording; these are the phrases the demo and the console use.

Value	Shown as
`allow`	Approved
`challenge`	Verify again
`hold`	Paused for review
`deny`	Blocked
`shadow`	Shadow mode (observing, not enforcing)
`sandbox`	Sandbox (test environment)

Generated from the API contract in contracts/ and the source in docs/. The build fails if this page disagrees with the running API.