speckit-ai-self-acceptance

To validate AI code changes before human acceptance, run a self-acceptance loop that gathers build, test, CDP, and runtime evidence to score against a rubric. This evaluates changes systematically and returns PASS, FAIL, or BLOCKED outcomes.

Rubric-based evidence collection for AI validation is the process of gathering build, test, CDP, log, and runtime data to score each rubric row. It records concrete blockers and produces PASS, FAIL, BLOCKED, or N/A outcomes to guide stage continuation decisions.

To collect CDP and runtime evidence for acceptance testing, trigger the self-acceptance loop for your current change set. The process automatically gathers logs, runtime data, and Chrome DevTools Protocol evidence based on your plan and selected gates.

Spec-kit self-acceptance works with Electron projects by gathering CDP evidence and runtime logs to evaluate changes. It requires a plan, acceptance rubric, gate evidence, and validation records to produce PASS, FAIL, or BLOCKED outcomes.

You should use automated acceptance testing for AI-driven code changes immediately after implementation to decide workflow continuation. It determines whether to proceed to human acceptance or return to fixes and revalidation based on evidence.

If a rubric row fails during AI self-acceptance, the evaluation records a FAIL or BLOCKED outcome and reports concrete blockers. This directs the workflow to return to fixes and revalidation rather than continuing to the next stage.