eval-criteria

Generates evaluation criteria from a software implementation plan into a scored eval file.

1|Updated Mar 7, 2026
One-click install
npx skills add https://github.com/ssv445/claude-skills --skill eval-criteria
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: eval-criteria
Source: https://github.com/ssv445/claude-skills/tree/main/eval-criteria
Command: npx skills add https://github.com/ssv445/claude-skills --skill eval-criteria

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

Generates task-specific evaluation criteria derived from implementation plans, replacing generic checklists with plan-derived pass/fail and quality criteria.

Core Features & Use Cases

  • Produces functional, code quality, UX, completeness, and taste criteria from a plan.
  • Automatically writes evals to docs/superpowers/evals and links to an implementation plan.
  • Integrates with eval-verifier to validate implementation after it completes.

Quick Start

Run /eval-criteria with a plan path to generate a plan-specific evaluation file.

Frequently Asked Questions about eval-criteria

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

FAQPage Schema
How do I generate evaluation criteria from a software implementation plan?

To generate evaluation criteria from a software implementation plan, run the eval-criteria command with your plan path. It parses goals, tasks, required files, tech stack, and personas to produce a scored eval file with functional and quality checks.

What is plan-derived task-specific eval criteria generation?

Plan-derived eval criteria generation replaces generic checklists by parsing a software implementation plan's goals, tasks, and personas to produce tailored pass/fail and quality metrics. It creates a scored evaluation file covering functional checks, code quality, and UX craftsmanship.

Can I automate plan analysis to create quality and completeness checks for different personas?

Yes, you can automate plan analysis to create quality and completeness checks across different personas. The tool parses implementation plans to tailor criteria, producing scored eval files that verify completeness and UX craftsmanship for each defined persona.

How do I verify implementation quality after completing a software plan?

To verify implementation quality after completing a software plan, the generated eval file integrates with eval-verifier. This validates the implementation against the plan-derived functional checks, code quality metrics, and taste-informed criteria.

Where are generated eval criteria files saved during plan analysis?

Generated eval criteria files are saved under the docs/superpowers/evals directory during plan analysis. The automation writes the scored eval file there and links it directly to the corresponding implementation plan.

Do I need any specific dependencies to generate eval criteria from my plans?

No specific dependencies are required to generate eval criteria from your plans. The tool operates independently to parse your implementation plan's tech stack, tasks, and goals, outputting the scored evaluation file directly to your docs folder.