skills.rest
GoogleChrome
GoogleChromeOFFICIAL
Explore all skills by this author →
GitHub

project-evals

Create standardized evaluation pipelines for web platform guidance using Playwright graders.

926|71|Updated Jan 27, 2026
Search Tags:#Playwright#grader calibration#web guidance#eval authoring#guidance testing#coding agent evals
One-click install
npx skills add https://github.com/GoogleChrome/modern-web-guidance-src --skill project-evals
Or copy as Structured Prompt for Agent▼
Please help me install this Agent Skill.
Skill: project-evals
Source: https://github.com/GoogleChrome/modern-web-guidance-src/tree/main/.agents/skills/project-evals
Command: npx skills add https://github.com/GoogleChrome/modern-web-guidance-src --skill project-evals

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve?

This skill solves the problem of unvalidated web guidance that may fail to produce correct, modern implementations when used by coding agents, ensuring guidance is reliable and effective before deployment.

Core Features & Use Cases

  • Eval File Authoring: Guides you to write compliant expectations.md, grader.ts, and tasks/task.md files aligned with eval pipeline requirements.
  • Grader Calibration: Helps you use the gd dev tool to generate, test, and fine-tune Playwright-based graders against reference and anti-example implementations.
  • Use Case: When authoring a new guidance for a modern web feature (e.g., Dialog API, View Transitions), use this skill to create evals that confirm coding agents implement the guidance correctly when given realistic developer prompts.

Quick Start

Use this skill to author and calibrate evaluation files for your new web guidance use case.

Frequently Asked Questions about project-evals

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

FAQPage Schema
How do I create evals to validate web platform guidance for coding agents?▼

To create evals for web platform guidance, you author standardized expectations.md, grader.ts, and task.md files that verify coding agents produce correct, modern implementations from your documentation before deployment.

What is grader calibration in Playwright for guidance testing?▼

Grader calibration uses the gd dev tool to generate, test, and fine-tune Playwright-based grader.ts files against reference and anti-example implementations to ensure automated grading accuracy for web platform features.

How do I test if coding agents correctly implement browser API and accessibility guidance?▼

You test coding agent implementations by building an evaluation pipeline that applies realistic developer prompts to your guidance, then automatically validates the generated outputs against reference and anti-example implementations.

Do I need Playwright to set up automated testing for coding agent evals?▼

Yes, Playwright is required for grader calibration. The eval pipeline uses Playwright-based grader files to automate testing of agent outputs against expected behavior for browser APIs, performance, and accessibility best practices.

When should I use an evaluation pipeline instead of manual code review for web guidance?▼

Use an evaluation pipeline when authoring new guidance for modern web features like the Dialog API or View Transitions, ensuring reliable, automated validation of agent outputs at scale before deployment.

Related Skills

Explore More Software Engineering Skills→
GoogleChromeGoogleChrome
926

investigating-eval-results

Diagnose root causes of low pass rates in AI coding agent web evaluation tests.

Official
Advanced
breethomasbreethomas
16

build-judge

Build a binary LLM evaluator that returns JSON Pass or Fail verdicts.

Community
Advanced
Nuwanda04Nuwanda04

senior-prompt-engineer

Analyze prompts, evaluate RAG systems, and design agentic workflows.

Community
Advanced
verselesverseles
179

desloppify

Scan codebases across 28 languages for technical debt metrics.

Official
Advanced