This page answers two questions: what prevents an invalid artifact from ever being created, and what detects one if it slips through anyway. Every mechanism listed is real and installed — nothing aspirational. Each ring below is read from the harness's own rules, not written for this page.
The founding observation: agents obey checkers, not documents. A rule that exists only as prose will eventually be violated by an honest agent under context pressure — proven live when a pull request satisfied exactly the sections the checker verified and dropped the two it didn't. So every contract rule must be a deterministic check, and every check must sit at the earliest chokepoint that can host it.
The model: three rings, by where a violation dies
| Ring | Where | What happens on violation | Who pays |
|---|---|---|---|
| Hooks | The agent's own machine (command, commit, push) | The action itself is refused — the invalid artifact never exists outside the agent's session. The exact errors feed back; the agent fixes and retries, in-session. | Nobody. Self-correcting. |
| Branch Rules | The forge (CI on every pull request) | The identical checks re-run; CI goes red; the merge gate makes red unmergeable by agents. Covers writers the local gates can't reach (web UI, humans, other tools). | Visible red — a human may look. Red on a gated rule now means a gate bug, not an agent failure. |
| Audits | After merge, continuously, across the whole forge | Drift is surfaced as findings, regardless of who or what wrote it — including history that predates the gates. | Scheduled clean-up, never a surprise mid-dispatch. |
The same check implementations run at ring 0 and ring 1 — one codebase, two enforcement points, so the local gates and CI can never disagree.