evidence-gate

What problem does it solve?

AI-driven missions require an objective, reproducible way to confirm that a worker's output satisfies the defined TaskContract. Manual reviews are error‑prone and cannot enforce consistent gating criteria across many tasks.

Core Features & Use Cases

• Three‑tiered gate: Precheck validates prerequisites, Mechanical runs eval commands, Contract + Rubric applies detailed acceptance, scope, risk, and scoring checks.
• Configurable gate profiles: Supports implementation changes, artifact‑only tasks, and closure‑ready cleanup with selective tier execution.
• Deterministic verdicts: Returns one of pass, fail, block, or error, integrates with retry budgets, escalation policies, and decision records.
• Use case: In an AI‑agent development pipeline, the evidence‑gate skill automatically validates code changes, documentation updates, or release tasks before they are marked verified.

Quick Start

Invoke the evidence‑gate skill on a specific task to automatically validate its output against the contract and receive a pass, fail, block, or error verdict.