Argus applies write-ahead audit, tenant isolation, permission, license evaluation, evidence and provenance validation, and a terminal audit record around every core-backed operation.
Argus itself does not generate investment judgments or execute orders. Client AI agents may use lawfully accessible facts and labeled source material for independent reasoning and output.
Governance boundary
- Every core-backed call records a start audit, checks tenant, permission, and license, validates schema, evidence, and provenance, and records a terminal audit.
- Investment intent does not block otherwise permitted facts or constrain the client Agent's independent output.
- Resolve every fact evidence id to source evidence in the same package.
- Compare known_time with the requested as_of cutoff.
- Preserve licenses and source-material labels in downstream systems.