skill-change-gate

What problem does it solve?

Identify and classify regressions introduced by a proposed change to an existing skill package, and return a structured decision without mutating the target.

Core Features & Use Cases

• Structural acceptance gating for skill changes, surfacing only gating results without mutating the target.
• Evidence classification and regression taxonomy (blocking, material, and warnings) to guide decisions.
• Supports multiple modes: candidate-gate, preflight-gate, post-validation-gate, and advisory-review.
• Outputs a standardized decision for downstream pipelines and integration with evaluator/benchmark workflows.
• Integrates with references and scripts to enable evidence collection and reproducible gating.

Quick Start

Run the gate on a candidate skill change by supplying the target folder and any before/after evidence to obtain a pass, pass-with-warnings, fail, or insufficient-evidence decision.