Quality gates
Quality gates turn verification results into an explicit accept-or-reject decision, instead of letting a run finish on unverified completion claims.
AgentDock Pro
Quality gates are part of AIgency, available only in AgentDock Pro.
Verification produces evidence and per-criterion pass/fail results. Quality gates are the policy layer that decides what those results actually mean for whether a task — or the run — is allowed to be considered done.
What a gate checks
- Whether every required acceptance criterion passed. A required criterion that fails or is blocked prevents the gate from passing, even if every other criterion succeeded.
- Whether any defects were found during verification that haven't been resolved — see Run history for the defect status lifecycle.
When a gate fails
A failed quality gate is what typically triggers a repair loop — the defect is recorded, and (within bounded limits) an agent attempts to fix it, followed by re-verification. A gate is never silently overridden by a completion message; if criteria genuinely aren't met, the task doesn't get marked done on the strength of the agent saying so.