TestingCodexClaude

LLM Eval Harness

Create lightweight evals for prompts and AI features.

Difficulty

Advanced

Risk

Medium

Helps with
Build eval set
Score outputs
Catch prompt regression
Use cases
AI QA
Prompt changes
Regression tests
Inside
SKILL.mdIncluded
ScriptsIncluded
ReferencesIncluded
AssetsNot listed
LLMevalsprompts
Checks
MediumScripts

Check before using

Review commands before execution.
Run on a branch or disposable workspace first.
Check file changes and external calls before shipping.
Source

Curated pack

Source

Similar skills

🧪Webapp Testing

Browser checks, screenshots, and interaction tests for local apps.

📱Mobile UI QA

Find cramped layouts, clipped text, and broken mobile flows.

🛒Checkout QA

Test checkout paths, edge cases, and payment-state copy.

🖼️Screenshot Diff Review

Compare UI screenshots and spot visual regressions.

🧪Playwright Tests

Generate focused browser tests for critical user paths.

🐞Bug Reproduction

Turn vague bug reports into repeatable repro steps.