gsd-eval-review

Audit executed AI phases against the AI-SPEC.md evaluation strategy and write EVAL-REVIEW.md.

1|Updated Sep 6, 2015
One-click install
npx skills add https://github.com/deerawan/dotfiles --skill gsd-eval-review-deerawan
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: gsd-eval-review
Source: https://github.com/deerawan/dotfiles/tree/main/claude/skills/gsd-eval-review
Command: npx skills add https://github.com/deerawan/dotfiles --skill gsd-eval-review-deerawan

SYSTEM DOCUMENTATION & REQUIREMENTS

💡 This Skill includes references (resource) components.

What problem does it solve?

It helps you identify whether a completed AI phase actually implemented the evaluation strategy you intended, so you can spot missing coverage and prevent blind spots from repeating.

Core Features & Use Cases

  • Evaluation coverage audit: Verifies whether the evaluation approach from AI-SPEC.md was implemented after the fact for a completed phase.
  • Actionable remediation planning: Produces an EVAL-REVIEW.md plan that includes a score, verdict, and specific gaps.
  • Phase-scoped review: Targets the phase specified by the phase number argument, defaulting to the last completed phase.
  • Use Case: You ran an AI workflow and later realize evaluations may not have been fully wired in; use this Skill to generate a remediation plan for what to change next run.

Quick Start

Run gsd-eval-review to audit the evaluation coverage for your last completed AI phase and generate EVAL-REVIEW.md with remediation steps.

Frequently Asked Questions about gsd-eval-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I audit AI evaluation coverage for a completed workflow phase?

You need AI evaluation coverage auditing when a completed AI workflow phase may not have fully implemented your intended evaluation strategy, risking blind spots. A retrospective review identifies missing coverage and prevents repeated gaps.

How do I generate a remediation plan for missing AI evaluation coverage?

Generate a remediation plan by auditing the executed AI phase against your intended evaluation strategy, which produces a review document containing a coverage score, a final verdict, identified gaps, and specific steps to fix the issues for the next run.

Can I target a specific phase number for a retrospective quality assurance review?

Yes, retrospective quality assurance review supports different task contexts and phase numbers across completed workflow phases. It targets the phase specified by the phase number argument, defaulting to the last completed phase if none is provided.

What is included in the evaluation coverage audit output?

An evaluation coverage audit outputs a review file containing a score, a verdict, identified gaps, and a remediation plan. This documents whether your intended evaluation strategy from your specification was actually implemented.

Does the evaluation review require a specific specification file to function?

The evaluation review requires reading an AI specification file containing your intended evaluation strategy to compare against the implemented evaluation behavior. It then writes a review file containing the score, verdict, gaps, and remediation plan.