Can I read an AI system's compliance status programmatically?

Everything the web app shows is available over the API, because the web app reads the same endpoints.

For monitoring or for an agent answering questions about your own estate, the useful ones are:

• GET /api/ai-systems — the inventory
• GET /api/risk-classification/{system_id}/latest — the current classification, including its citations, the verifier’s confidence score and any unsupported claims
• GET /api/compliance/systems/{system_id} — per-system obligation status
• GET /api/compliance/dashboard — the roll-up across systems
• GET /api/compliance/urgent-actions — what is time-critical
• GET /api/audit — the organisation’s audit log

The classification response is worth reading in full rather than reducing to a tier string. verifier_confidence, unsupported_claims and the requires-expert-review flag are the difference between “this system is high-risk” and “this system is high-risk and we are confident about why” — and an agent that drops those fields is discarding exactly the information that makes the answer safe to act on.