evals-skills

evals-skills

Evaluate skills for AI coding agents to help build LLM evaluations.

Agent SkillDevOpen source
Type
Agent Skill
Open source
Yes
GitHub Stars
★ 1.4k
Source
skill-github

Overview

These skills are designed to guide AI coding agents and help users build LLM evaluations. They can help avoid common mistakes and recommend other skills to resolve issues. Suitable for engineers and product managers, especially those new to evaluation. After installing the plugin via Claude Code or Skills CLI, use the command `/evals-skills:<skill-name>` to invoke various skills.

Capabilities

  • Audit evaluation pipelines and identify priority issues
  • Guide users to review traces and categorize failures
  • Generate diverse synthetic test inputs
  • Design evaluators using LLMs as judges
  • Calibrate LLM judges against human labels
  • Evaluate retrieval and generation quality in RAG pipelines

Use cases

Audit and improve LLM evaluation pipelinesGenerate synthetic data to test modelsDesign evaluators for subjective quality criteriaCalibrate model consistency with human labels

Setup

Requires: API KeyNode 环境
Run the following commands in Claude Code: "/plugin marketplace add hamelsmu/evals-skills" and "/plugin install evals-skills@hamelsmu-evals-skills"; or use npx: "npx skills add https://github.com/hamelsmu/evals-skills"

This information was compiled by AI from public sources and may contain inaccuracies — please refer to the source.

FAQ

How do I get started with these skills?

Start with the `eval-audit` skill and follow the steps in the README to install and run it.

Who are these skills for?

Suitable for engineers and product managers, especially those just getting started with evaluation.

Related skills