Policy

Evidence model

The evidence model focuses on observable agent actions and boundary crossings, then preserves integrity through canonical hashes, signed checkpoints, and verifier-friendly exports.

01

Capture

Native agent streams, prompts where available, tool calls, shell commands, process exits, file changes, and git diffs.

02

Seal

Canonical event encoding, stable payload hashes, hash-chain links, manifest hashes, and local verification metadata.

03

Stream

Remote audit writer checkpoints sequence the received stream and produce server-owned signed evidence state.

04

Verify

Evidence packages can be inspected independently to check hashes, signatures, timestamps, custody, and run boundaries.