Evals and Benchmarks skills for Claude Code and Codex
Skills for measuring AI quality: LLM-as-judge setups, eval-driven development, agent evaluation systems, regression suites for skills and prompts, Phoenix and Arize evaluators, and benchmark harnesses such as lm-evaluation-harness, NeMo Evaluator and Hugging Face community evals. They give an agent a way to prove a model, prompt or skill improved instead of asserting it.
Highest-scoring eval skills in the October 22, 2026 snapshot: advanced-evaluation from muratcankoylan/Agent-Skills-for-Context-Engineering (86/100), huggingface-community-evals from huggingface/skills (86/100) and skill-creator from anthropics/skills (83/100). Scores come from the Suede Skill Check.
- 27 listed skills
- 15 repositories
- 6 with full pages
13 from official publishers. 7 bundle scripts. 2 connect to an outside service. Listed in AI Agents and LLM Engineering (26), Testing and QA (1).
All 27 eval skills
No matching skills. Try fewer words or clear your search.
| Skill | Repository | What it does |
|---|---|---|
| advanced-evaluation 86/100 · B | muratcankoylan/Agent-Skills-for-Context-Engineering | Guides advanced LLM evaluation, including LLM-as-judge systems, direct scoring, pairwise comparison, rubric calibration, evaluator bias mitigation, confidence scoring, and automated quality assessment. |
| evaluation 86/100 · B | muratcankoylan/Agent-Skills-for-Context-Engineering | Guides building agent evaluation systems with deterministic checks, regression suites, multi-dimensional rubrics, quality gates, production monitoring, baseline comparison, and outcome measurement for agent pipelines. |
| huggingface-community-evals 86/100 · B | huggingface/skills | Runs evaluations for Hugging Face Hub models on local hardware using inspect-ai and lighteval, including backend selection and choosing between vLLM, Transformers, and accelerate. |
| prompt-optimizer 81/100 · B | getsentry/skills | Creates and refines agent, system, and developer prompts and reusable prompt templates, including porting prompts between OpenAI, Claude, and Gemini and building prompt evals. |
| skill-creator 83/100 · B | anthropics/skills | Guides creating new skills, improving existing ones, running evals, benchmarking skill performance with variance analysis, and optimizing a skill's description for better triggering. |
| test-skill 78/100 · C | NeoLabHQ/context-engineering-kit | Tests skills before deployment by running a baseline without the skill, writing fixes for failures, and iterating to close loopholes under pressure. |
| agent-evaluation 64/100 · D | NeoLabHQ/context-engineering-kit | Evaluate and improve Claude Code commands, skills, and agents. Use when testing prompt effectiveness, validating context engineering choices, or… |
| agent-platform-eval-flywheel 74/100 · C | google/skills | Measures and improves the quality of AI models and agents on Google Cloud using the Eval Quality Flywheel methodology. Use when generating synthetic… |
| agent-usability-test 63/100 · D | serpapi/skills | Test whether agents can discover and use your tool — not whether agents are capable. The subject under test is the interface. A bad score means fix… |
| agentic-eval 77/100 · C | github/awesome-copilot | Patterns and techniques for evaluating and improving AI agent outputs. Use this skill when: - Implementing self-critique and reflection loops… |
| arize-evaluator 68/100 · D | github/awesome-copilot | Handles LLM-as-judge evaluation workflows on Arize including creating/updating evaluators, running evaluations on spans or experiments, managing… |
| digital-health-clinical-asr-eval 79/100 · C | NVIDIA/skills | Stage 3 of Clinical ASR Flywheel. Score a NeMo manifest, produce the five-section KER leaderboard (by-ipa_source diagnostic). Not for ASR auth… |
| eval-driven-dev 82/100 · B | github/awesome-copilot | Improve AI application with evaluation-driven development. Define eval criteria, instrument the application, build golden datasets, observe and… |
| eval-engineering 82/100 · B | langchain-ai/langchain-skills | Inspect an agent repository and optional traces, interview the user, write reviewed Task Specs, build and audit Harbor tasks, and bootstrap reusable… |
| evaluating-code-models 69/100 · D | Orchestra-Research/AI-Research-SKILLs | Evaluates code generation models across HumanEval, MBPP, MultiPL-E, and 15+ benchmarks with pass@k metrics. Use when benchmarking code models… |
| evaluating-cosmos-policy 69/100 · D | Orchestra-Research/AI-Research-SKILLs | Evaluates NVIDIA Cosmos Policy on LIBERO and RoboCasa simulation environments. Use when setting up cosmos-policy for robot manipulation evaluation… |
| evaluating-llms-harness 69/100 · D | Orchestra-Research/AI-Research-SKILLs | Evaluates LLMs across 60+ academic benchmarks (MMLU, HumanEval, GSM8K, TruthfulQA, HellaSwag). Use when benchmarking model quality, comparing… |
| judge 66/100 · D | NeoLabHQ/context-engineering-kit | Launch a meta-judge then a judge sub-agent to evaluate results produced in the current conversation |
| judge-with-debate 66/100 · D | NeoLabHQ/context-engineering-kit | Evaluate solutions through multi-round debate between independent judges until consensus |
| nemo-evaluator-sdk 69/100 · D | Orchestra-Research/AI-Research-SKILLs | Evaluates LLMs across 100+ benchmarks from 18+ harnesses (MMLU, HumanEval, GSM8K, safety, VLM) with multi-backend execution. Use when needing… |
| phoenix-evals 67/100 · D | github/awesome-copilot | Build and run evaluators for AI/LLM applications using Phoenix. |
| prompt-engineer 84/100 · B | jeffallan/claude-skills | Writes, refactors, and evaluates prompts for LLMs — generating optimized prompt templates, structured output schemas, evaluation rubrics, and test… |
| rag-eval 81/100 · B | NVIDIA/skills | Filesystem RAG benchmarks: corpus/, train.json, evaluate_rag.py (RAGAS quality). Not for prod monitoring, latency/throughput benchmarking (use… |
| rook-skill 75/100 · C | LambdaTest/agent-skills | Use rook to test, evaluate, red-team or regression-check an AI agent the user owns, or to interpret saved Rook results. Does not apply to ordinary… |
| skill-evaluator 84/100 · B | HeshamFS/materials-simulation-skills | Rigorously evaluate an Agent Skill end-to-end across ANY coding-agent CLI — verify its scripts emit the documented numbers (deterministic checks)… |
| Suede AI Eval Skills 91/100 · A | JasonColapietro/suede-creator-skills | Suede AI Eval Skills. Use when asked to "write evals for this AI feature" or assess LLM, RAG, agent, classifier or generated-media behavior… |
| test-prompt 70/100 · C | NeoLabHQ/context-engineering-kit | Use when creating or editing any prompt (commands, hooks, skills, subagent instructions) to verify it produces desired behavior - applies… |
How to choose eval skills
Decide what you are scoring. Benchmark harnesses (evaluating-llms-harness, nemo-evaluator-sdk, evaluating-code-models) rank models on public datasets; eval-driven development and agent evaluation skills build test sets from your own application.
Platform evaluators on Arize, Phoenix and Google's Agent Platform store results in that service and read its API key. Local harnesses run on your hardware and may need a GPU, which the bundled scripts list shows.
For skills and prompts, use an eval that compares against a baseline. test-skill and skill-creator run the task with and without the change, so you see what the change actually did.
Ranked eval shortlists
Related topics
Prompt Engineering 28Prompt Generators 19Skill Authoring 37Agent Tracing and Observability 41Subagents and Multi-Agent Orchestration 40Claude Code Setup and Hooks 31Codex, Cursor and Coding Agents 23Inference and Model Serving 39
Find your next useful skill
Explore Rising Skills with quality scores and source reports, or search from inside your agent.
Get Rising Skills by email · Read Rising Skills · Install the finder · State of Agent Skills