skill-change-gate

A change gate evaluates candidate diffs or patches against before and after skill packages to identify and classify regressions in activation, scope, safety, or packaging, returning a structured pass or fail decision without mutating the target.

Skill change gating is a structural acceptance process that classifies regressions from proposed skill package modifications using candidate evidence like diffs and validator outputs, returning a standardized final decision of pass, pass-with-warnings, fail, or insufficient-evidence.

Yes, the gate evaluates hardening patches and manual edits by analyzing the target folder and before or after evidence to identify blocking regressions in local references, validation, and output contracts without modifying the original package.

Running a preflight gate requires candidate evidence including a diff or patch, before and after packages, and any validator or benchmark outputs to accurately assess output contracts and packaging regressions for a pass or fail decision.

Use advisory review mode when you need to surface material regressions and warnings from a skill package change without enforcing a strict pass, pass-with-warnings, fail, or insufficient-evidence blocking decision in your downstream pipelines.

The gate returns an insufficient-evidence decision when provided candidate evidence like diffs, before and after packages, or validator outputs is inadequate to confidently classify activation, validation, or packaging regressions.