rahulchaube1/-skills-for-agents/evaluation-harness-design

Claude Code and agent skill by rahulchaube1/-skills-for-agents

Trust tier 2
  • community
LLM review pending
Source check pending

Checking that the SKILL.md file is still in the repository.

Designs repeatable evaluation harnesses for AI and software workflows. Use when you need to measure quality, compare approaches, or prevent regressions from landing.

Install

npx vskill@latest install rahulchaube1/-skills-for-agents/evaluation-harness-design

SKILL.md

The same SKILL.md was published earlier as rahulchaube1/-skills-for-agents/evaluation-harness-design.

[!NOTE]
Skills for Agents card
Every skill in this repo now follows a cleaner, card-like reading experience.
Keep the workflow focused: Overview → When to Use → Process → Verification.
[!NOTE]
Evaluation Harness Design
Create stable scoring and comparison loops.
Workflow lens: Overview → When to Use → Process → Verification

Evaluation Harness Design

Overview

Evaluation Harness Design is a focused workflow for handling this class of work with more clarity and less rework.

When to Use

  • The task fits the evaluation harness design workflow.
  • You need a repeatable process rather than an ad hoc answer.
  • You want evidence, not just a plausible result.

When NOT to use: Pure curiosity, unrelated cleanup, or a task that belongs to a different phase.

Process

  1. Clarify the target outcome.
  2. Gather the minimum context needed to act.
  3. Apply the workflow in small, verifiable steps.
  4. Verify with concrete evidence before moving on.

Common Rationalizations

RationalizationReality
"This is too small to need a process."Small changes still create hidden regressions.
"I can verify later."Verification is cheaper before merge than after release.
"The obvious path is probably fine."Explicit assumptions beat silent guesses.

Red Flags

  • The task is moving forward without a clear outcome.
  • The workflow is being compressed into a single unverified step.
  • Output is being judged by appearance instead of evidence.

Verification

  • [ ] The result matches the requested outcome.
  • [ ] Assumptions are documented if anything was ambiguous.
  • [ ] The change can be reviewed or rerun by someone else.

Security

  • Security scanPassed, 100 of 100Pattern checks on the skill's files.
  • Model reviewNot runA model review is scheduled for the next rescan.

Quality evaluation

  • Task evaluationNot runNo evaluation yet. Run one in Skill Studio.

Source

Popularity

Repository stars
2
7-day trend score
10

Stars and forks are for the whole GitHub repository, not this skill alone. Not refreshed yet.

Works with

Any agent vskill installs to: 56 targets in all. These 39 have a page on this site.

  • Claude Code
  • Cursor
  • GitHub Copilot
  • Windsurf
  • Codex
  • Gemini CLI
  • Cline
  • Amp
  • Roo Code
  • Goose
  • Continue
  • Kilo Code
  • Junie
  • Trae
  • OpenHands
  • Qwen Code
  • AI Agent
  • Tabnine Chat
  • Sourcegraph Cody
  • Amazon Q
  • Firebase Genkit
  • Sweep
  • Devin
  • Phind
  • Mentat
  • Claude Desktop
  • Aider
  • GPT Pilot
  • Plandex
  • bolt.new
  • v0
  • Replit Agent
  • Anse
  • TypingMind
  • BetterChatGPT
  • LibreChat
  • Warp AI
  • Pieces
  • Fig AI
Embed badge
vskill VERIFIED badge
[![vskill VERIFIED](https://verified-skill.com/api/v1/skills/rahulchaube1/-skills-for-agents/evaluation-harness-design/badge)](https://verified-skill.com/skills/rahulchaube1/-skills-for-agents/evaluation-harness-design)

Something wrong with this skill or its page? Tell us.