What problem does it solve? Autonomous task workflows need an independent quality gate that decides whether completed work actually meets the original request, without re-running tests or expanding scope. This Skill acts as the single judge at the closing gate of every yolo task and at the planning gate of critical tasks, keeping human approval focused where it matters. ## Core Features & Use Cases - Gate-based judgment: Operates at the 005_closing gate of every yolo task and the 003 gate of critical tasks, always in a fresh context separate from planners and executors. - Materiality-filtered findings: Every objection must pass a five-question materiality test with verifiable evidence, severity, and remedy before it can block a task. - Structured verdicts and deltas: Produces a .verify.md verdict with machine-readable markers and, on returns, a delta classifying the defect as directed repair, execution failure, plan gap, or wrong premise. - Use Case: An orchestrator finishes an autonomous coding task and invokes this Skill to audit the integrated diff against the card's objective; the judge approves with a memory entry or returns the task to the correct stage with a precise delta. ## Quick Start Ask the agent to act as judge-dredd and judge the task at its 005_closing gate by reading the card, plan, and integrated diff, then write the verdict in the .verify.md file.
