Gate the complete insight against 6 quality criteria; FAIL halts pipeline, REVISE triggers targeted re-run
insight_validationfullInsight (JSON string aggregating all outputs from Steps 1-6) from insight tacticinsight_validation({ fullInsight }) via dare-agentsrevisionGuidance.fromStepgates[] — per-gate assessment: name, status (PASS|FAIL), rationaleoverallVerdict — PASS (all gates pass), FAIL (structural failure, cannot revise), REVISE (targeted fix possible)failedGates[] — names of gates that did not passrevisionGuidance — present only when verdict is REVISE; specifies fromStep (1-6) and what to fix