LLM Benchmarking with lm-evaluation-harness
Orchestra-Research/AI-Research-SKILLs
Runs lm-evaluation-harness to benchmark language models on academic suites such as MMLU, GSM8K and HumanEval, compare models and track training checkpoints.
适用于 Claude Code 会话的正规评测框架（Evaluation Framework），实现了评测驱动开发（Eval-Driven Development, EDD）原则
Install Eval Harness
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a claude-codeProject install by default; add -g for ~/.claude/skills/.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harness --agent claude-codeProject scope by default; add --scope user for a personal install. Needs GitHub CLI 2.90.0 or later (public preview).
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .claude/skills && cp -r skills-src/.agents/skills/eval-harness .claude/skills/eval-harness && rm -rf skills-srcUse ~/.claude/skills/ instead of .claude/skills for a personal install. The folder must contain SKILL.md.
Claude Code skills documentation · loads skills from .claude/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .claude/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.Claude Code copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$skill-installer install https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harnessType this inside Codex. $skill-installer <name> installs a curated skill from openai/skills. The installer writes to $CODEX_HOME/skills (default ~/.codex/skills). Restart Codex if the skill does not show up.
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a codexProject install goes to .agents/skills/; add -g for ~/.codex/skills/.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harness --agent codexProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .agents/skills && cp -r skills-src/.agents/skills/eval-harness .agents/skills/eval-harness && rm -rf skills-srcUse ~/.agents/skills/ instead of .agents/skills for a personal install.
Codex skills documentation · loads skills from .agents/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .agents/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.Codex copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a cursorProject install goes to .agents/skills/; add -g for ~/.cursor/skills/.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harness --agent cursorProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .cursor/skills && cp -r skills-src/.agents/skills/eval-harness .cursor/skills/eval-harness && rm -rf skills-srcUse ~/.cursor/skills/ instead of .cursor/skills for a personal install.
Cursor skills documentation · loads skills from .cursor/skills/, .agents/skills/, .claude/skills/, .codex/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .cursor/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.Cursor copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gemini skills install https://github.com/xu-xiang/everything-claude-code-zh.git --path .agents/skills/eval-harness--scope user (default) or --scope workspace; --path is the subfolder of the repo that holds the skill; --consent skips the security confirmation prompt.
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a gemini-cliProject install goes to .agents/skills/; add -g for ~/.gemini/skills/.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harness --agent gemini-cliProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .gemini/skills && cp -r skills-src/.agents/skills/eval-harness .gemini/skills/eval-harness && rm -rf skills-srcUse ~/.gemini/skills/ instead of .gemini/skills for a personal install, then run /skills reload.
Gemini CLI skills documentation · loads skills from .gemini/skills/, .agents/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .gemini/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.Gemini CLI copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harnessInstalls for Copilot at project scope by default; add --scope user for a personal install. Preview a skill first with gh skill preview. Needs GitHub CLI 2.90.0 or later (public preview).
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a github-copilotProject install goes to .agents/skills/; add -g for ~/.copilot/skills/.
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .github/skills && cp -r skills-src/.agents/skills/eval-harness .github/skills/eval-harness && rm -rf skills-srcUse ~/.copilot/skills/ instead of .github/skills for a personal install. Commit .github/skills so cloud agent and code review can use it.
GitHub Copilot skills documentation · loads skills from .github/skills/, .claude/skills/, .agents/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .github/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.GitHub Copilot copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
$ npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a opencodeOpenCode documents no install command of its own. Project install goes to .agents/skills/; add -g for ~/.config/opencode/skills/.
$ gh skill install xu-xiang/everything-claude-code-zh eval-harness --agent opencodeProject scope by default (.agents/skills/); add --scope user for a personal install.
$ git clone --depth 1 https://github.com/xu-xiang/everything-claude-code-zh.git skills-src && mkdir -p .opencode/skills && cp -r skills-src/.agents/skills/eval-harness .opencode/skills/eval-harness && rm -rf skills-srcUse ~/.config/opencode/skills/ instead of .opencode/skills for a personal install.
OpenCode skills documentation · loads skills from .opencode/skills/, .claude/skills/, .agents/skills/
Install the "eval-harness" agent skill from https://github.com/xu-xiang/everything-claude-code-zh/tree/main/.agents/skills/eval-harness into .opencode/skills/eval-harness/ in this project. Copy the whole folder (SKILL.md and every file beside it), keep the folder name "eval-harness", then confirm the skill loads.OpenCode copies the folder itself, the same result as the manual copy. Check what it changed before you commit it.
Facts
- Skill name
eval-harness- GitHub stars
- 2k
- Token cost
- ~974 tokens
- SKILL.md length
- 129 words
- Files
- 2
- Skills in repo
- 78
- Repo updated
- First seen
- Licence
- MIT
At a glance
适用于 Claude Code 会话的正规评测框架（Evaluation Framework），实现了评测驱动开发（Eval-Driven Development, EDD）原则
- Works in 7 steps: 基于代码的评分器（Code-Based Grader） → 基于模型的评分器（Model-Based Grader） → 人工评分器（Human Grader） → …
- Tasks that involve LLM evaluation
- SKILL.md covers 何时激活, 核心理念, 评测类型 and 评分器（Grader）类型, plus 6 more sections
- Calls npm
What it does
Eval Harness is an agent skill from xu-xiang/everything-claude-code-zh. 适用于 Claude Code 会话的正规评测框架（Evaluation Framework），实现了评测驱动开发（Eval-Driven Development, EDD）原则
Its SKILL.md is about 970 tokens, which your agent loads only when the skill is triggered. The skill folder holds 2 other files (for example `agents/openai.yaml`).
It sits in AI & LLM Engineering, covering LLM evaluation. The repository describes itself as: everything-claude-code 中文翻译项目：完整的 Claude Code 配置集合（agents, skills, hooks, commands, rules, MCPs）。源自 Anthropic 黑客松获胜者的实战配置，助力中文工程师高效理解与使用 Claude Code。 The licence is MIT.
When your agent uses it
- Tasks that involve LLM evaluation
Example prompts
- “/eval-harness”
Workflow steps
7 steps, taken from the step headings in SKILL.md.
What it can do on your machine
Read from SKILL.md and the folder at commit dfbf946. It shows what the files ask for, not the result of running them.
Tool permissions
Pre-approves nothing: there is no allowed-tools line, so your agent's usual permission prompts apply.
From allowed-tools in the SKILL.md frontmatter.
Runs code
Shell commands in SKILL.md call:
npm
From the folder's file list and the shell code blocks in SKILL.md.
Network
No URLs in SKILL.md. Its commands use npm, which can reach the network depending on how they are called.
From URLs in SKILL.md, links to its own repository left out.
Credentials
Names no API keys, tokens, secrets or passwords.
From names ending in _API_KEY, _TOKEN, _SECRET, _KEY or _PASSWORD in SKILL.md.
Context cost
Eval Harness loads about 974 tokens when it runs. Until then it costs about 26 tokens; SKILL.md has 129 words of instructions outside code blocks.
Estimates: characters ÷ 4, the usual rule of thumb; real counts depend on the model's tokenizer. Scripts and assets cost tokens only if the agent reads them.
Safety
Auto-check passed
The automated check found no risky patterns in SKILL.md.
Automated static check — not a guarantee. Review scripts before installing. It scans the text of SKILL.md for risky patterns (piping downloads into a shell, reading credential files, hidden Unicode, destructive commands); files beside SKILL.md are not scanned.
SKILL.md
The full file from xu-xiang/everything-claude-code-zh at commit dfbf946, republished under its MIT licence (© xu-xiang). 129 words, ~974 tokens.
Download SKILL.mdSave it as
.claude/skills/eval-harness/SKILL.md (or your agent's skills folder). This skill also uses 1 other file; get the full folder from GitHub.- name
- eval-harness
- description
- 适用于 Claude Code 会话的正规评测框架（Evaluation Framework），实现了评测驱动开发（Eval-Driven Development, EDD）原则
- origin
- ECC
- tools
- Read, Write, Edit, Bash, Grep, Glob
评测框架（Eval Harness）技能（Skill）
一个用于 Claude Code 会话的正规评测框架（Evaluation Framework），旨在落实评测驱动开发（Eval-Driven Development, EDD）原则。
何时激活
- 为 AI 辅助工作流设置评测驱动开发（EDD）
- 为 Claude Code 任务的完成情况定义通过/失败标准
- 使用 pass@k 指标衡量智能体（Agent）的可靠性
- 为提示词（Prompt）或智能体（Agent）的变更创建回归测试套件
- 跨模型版本对智能体（Agent）性能进行基准测试
核心理念
评测驱动开发（Eval-Driven Development）将评测（Eval）视为“AI 开发中的单元测试”：
- 在实现之前定义预期行为
- 在开发过程中持续运行评测（Evals）
- 跟踪每次变更带来的回归（Regressions）
- 使用 pass@k 指标进行可靠性度量
评测类型
能力评测（Capability Evals）
测试 Claude 是否能够完成其之前无法完成的任务：
[CAPABILITY EVAL: feature-name]
Task: 描述 Claude 应该完成的任务
Success Criteria:
- [ ] 准则 1
- [ ] 准则 2
- [ ] 准则 3
Expected Output: 预期结果的描述回归评测（Regression Evals）
确保变更不会破坏现有功能：
[REGRESSION EVAL: feature-name]
Baseline: SHA 或检查点（checkpoint）名称
Tests:
- existing-test-1: PASS/FAIL
- existing-test-2: PASS/FAIL
- existing-test-3: PASS/FAIL
Result: X/Y 通过 (之前为 Y/Y)评分器（Grader）类型
1. 基于代码的评分器（Code-Based Grader）
使用代码进行确定性检查：
# 检查文件是否包含预期模式
grep -q "export function handleAuth" src/auth.ts && echo "PASS" || echo "FAIL"
# 检查测试是否通过
npm test -- --testPathPattern="auth" && echo "PASS" || echo "FAIL"
# 检查构建是否成功
npm run build && echo "PASS" || echo "FAIL"2. 基于模型的评分器（Model-Based Grader）
使用 Claude 对开放式输出进行评估：
[MODEL GRADER PROMPT]
评估以下代码变更：
1. 它是否解决了所述问题？
2. 结构是否良好？
3. 是否处理了边缘情况？
4. 错误处理是否恰当？
Score: 1-5 (1=差, 5=优秀)
Reasoning: [解释]3. 人工评分器（Human Grader）
标记以供人工复核：
[HUMAN REVIEW REQUIRED]
Change: 变更内容描述
Reason: 为何需要人工复核
Risk Level: LOW/MEDIUM/HIGH指标（Metrics）
pass@k
“在 k 次尝试中至少成功一次”
- pass@1: 首次尝试成功率
- pass@3: 3 次尝试内的成功率
- 典型目标：pass@3 > 90%
pass^k
“所有 k 次试验均成功”
- 更高的可靠性门槛
- pass^3: 连续 3 次成功
- 用于关键路径（Critical Paths）
评测工作流（Eval Workflow）
1. 定义（编码前）
## EVAL DEFINITION: feature-xyz
### 能力评测（Capability Evals）
1. 能够创建新用户账号
2. 能够验证邮箱格式
3. 能够安全地哈希密码
### 回归评测（Regression Evals）
1. 现有登录功能仍然正常
2. 会话管理未改变
3. 注销流程完好无损
### 成功指标
- 能力评测的 pass@3 > 90%
- 回归评测的 pass^3 = 100%2. 实现
编写代码以通过定义的评测（Evals）。
3. 评测
# 运行能力评测
[运行每个能力评测，记录 PASS/FAIL]
# 运行回归评测
npm test -- --testPathPattern="existing"
# 生成报告4. 报告
EVAL REPORT: feature-xyz
========================
Capability Evals:
create-user: PASS (pass@1)
validate-email: PASS (pass@2)
hash-password: PASS (pass@1)
Overall: 3/3 passed
Regression Evals:
login-flow: PASS
session-mgmt: PASS
logout-flow: PASS
Overall: 3/3 passed
Metrics:
pass@1: 67% (2/3)
pass@3: 100% (3/3)
Status: READY FOR REVIEW集成模式（Integration Patterns）
实现前
/eval define feature-name在 .claude/evals/feature-name.md 创建评测定义文件
实现中
/eval check feature-name运行当前评测并报告状态
实现后
/eval report feature-name生成完整的评测报告
评测存储（Eval Storage）
在项目中存储评测（Evals）：
.claude/
evals/
feature-xyz.md # 评测定义
feature-xyz.log # 评测运行历史
baseline.json # 回归基准最佳实践
- 在编码之前定义评测 - 强制对成功标准进行清晰思考
- 频繁运行评测 - 尽早发现回归问题
- 长期跟踪 pass@k - 监控可靠性趋势
- 尽可能使用代码评分器 - 确定性 > 概率性
- 安全相关的由人工复核 - 绝不要完全自动化安全检查
- 保持评测速度快 - 慢的评测不会被经常运行
- 评测与代码版本同步 - 评测是一等公民（First-class Artifacts）
示例：添加身份验证（Authentication）
## EVAL: add-authentication
### 阶段 1：定义 (10 分钟)
能力评测：
- [ ] 用户可以使用邮箱/密码注册
- [ ] 用户可以使用有效凭据登录
- [ ] 无效凭据被拒绝并返回正确错误
- [ ] 会话在页面重新加载后保持
- [ ] 注销会清除会话
回归评测：
- [ ] 公共路由仍然可以访问
- [ ] API 响应未改变
- [ ] 数据库架构兼容
### 阶段 2：实现 (时间视情况而定)
[编写代码]
### 阶段 3：评测
运行：/eval check add-authentication
### 阶段 4：报告
EVAL REPORT: add-authentication
==============================
Capability: 5/5 passed (pass@3: 100%)
Regression: 3/3 passed (pass^3: 100%)
Status: SHIP IT (可以发布)© xu-xiang, MIT. Rendered from Markdown: HTML in the file is shown as text, images as links, and headings moved down two levels. Raw file
Files
SKILL.md and 1 other file in .agents/skills/eval-harness of xu-xiang/everything-claude-code-zh.
- SKILL.md
- agents/openai.yaml
Open the folder on GitHubat commit dfbf946
Compare with similar skills
Eval Harness next to the 5 skills that share the most tags, products or categories with it. Stars are the repository's; “used in” counts other GitHub owners with a copy.
| Skill | Stars | Used in | Tokens | Auto-check | Licence | Repo updated |
|---|---|---|---|---|---|---|
| Eval Harness this skillxu-xiang/everything-claude-code-zh | 2k | — | ~974 | Automated safety check: Pass | MIT | |
| LLM Benchmarking with lm-evaluation-harnessOrchestra-Research/AI-Research-SKILLs | 13k | 8 repos | ~3k | Automated safety check: Pass | MIT | |
| Hugging Face Local Model Evalshuggingface/skills | 11k | 2 repos | ~1.6k | Automated safety check: Pass | Apache-2.0 | |
| Looperksimback/looper | 710 | — | ~2.7k | Automated safety check: Notes | MIT | |
| Agent Eval Engineeringlangchain-ai/langchain-skills | 1.3k | — | ~4k | Automated safety check: Pass | MIT | |
| Quality FlywheelGoogleCloudPlatform/vertex-ai-samples | 792 | — | ~2k | Automated safety check: Pass | Apache-2.0 |
Similar skills
- Official
Hugging Face Local Model Evals
huggingface/skills
Runs evaluations of Hugging Face Hub models on local hardware with inspect-ai or lighteval, and helps choose between vLLM, Transformers and accelerate backends.
Looper
ksimback/looper
Scaffold a well-designed agent loop with best-practice coaching and a cross-model review council.
- Official
Agent Eval Engineering
langchain-ai/langchain-skills
Builds agent evaluations in stages: inspect the repository and traces, agree a Task Spec with you, then build, audit and run a Harbor task with an independent verifier.
Quality Flywheel
GoogleCloudPlatform/vertex-ai-samples
Evaluate and improve GenAI models and agents using the Google GenAI Evaluation SDK.
Eval Harness
cloudnative-co/claude-code-starter-kit
Formal evaluation framework for Claude Code sessions implementing eval-driven development (EDD) principles.
More from xu-xiang/everything-claude-code-zh
All 78 skills in this repoConfigure Ecc
xu-xiang/everything-claude-code-zh
Everything Claude Code 的交互式安装程序 — 引导用户选择并安装技能和规则到用户级或项目级目录，验证路径，并可选择优化已安装文件。
Continuous Learning V2
xu-xiang/everything-claude-code-zh
基于本能（Instinct）的学习系统，通过钩子（hooks）观察会话，创建带有置信度评分的原子本能，并将其演化为技能（Skills）、命令（Commands）或智能体（Agents）。v2.1 版本增加了项目作用域（project-scoped）的本能，以防止跨项目污染。
API Design
xu-xiang/everything-claude-code-zh
生产级 API 的 REST API 设计模式，包括资源命名、状态码、分页、过滤、错误响应、版本控制和速率限制. An agent skill from xu-xiang/everything-claude-code-zh.
Backend Patterns
xu-xiang/everything-claude-code-zh
后端架构模式、API 设计、数据库优化以及适用于 Node.js、Express 和 Next.js API 路由的服务端最佳实践。
Backend Patterns
xu-xiang/everything-claude-code-zh
后端架构模式、API 设计、数据库优化以及 Node.js、Express 和 Next.js API 路由的服务端最佳实践。
Backend Patterns
xu-xiang/everything-claude-code-zh
后端架构模式、API 设计、数据库优化以及针对 Node.js、Express 和 Next.js API 路由的服务端最佳实践。
Related
Categories
Questions about Eval Harness
What does Eval Harness do?
适用于 Claude Code 会话的正规评测框架（Evaluation Framework），实现了评测驱动开发（Eval-Driven Development, EDD）原则. Eval Harness is an agent skill from xu-xiang/everything-claude-code-zh.
When should I use Eval Harness?
Eval Harness fits situations like: tasks that involve LLM evaluation.
How do I install Eval Harness in Claude Code?
Run `npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a claude-code`. Or copy the skill folder (.agents/skills/eval-harness in xu-xiang/everything-claude-code-zh) into .claude/skills/eval-harness in your project. Claude Code loads it when a task matches its description.
How do I install Eval Harness in Codex?
Run `npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a codex`. Or copy the skill folder (.agents/skills/eval-harness in xu-xiang/everything-claude-code-zh) into .agents/skills/eval-harness in your project. Codex loads it when a task matches its description.
Can I use Eval Harness in Cursor, Gemini CLI or GitHub Copilot?
Cursor, Gemini CLI, GitHub Copilot and OpenCode also load SKILL.md folders. With the skills CLI, run `npx skills add xu-xiang/everything-claude-code-zh --skill eval-harness -a cursor` (or -a gemini-cli, github-copilot or opencode for the others). To copy it by hand, put the folder in .cursor/skills/eval-harness, .gemini/skills/eval-harness, .github/skills/eval-harness and .opencode/skills/eval-harness in your project.
What does Eval Harness need to run?
Going by SKILL.md and its folder, Eval Harness needs the command-line tools its instructions call (npm).
Does Eval Harness access the network?
SKILL.md contains no URLs. Its commands use npm, which can reach the network depending on how they are called. This is read from the text; nothing was executed.
Is Eval Harness safe to install?
Our automated static check of SKILL.md found no risky patterns, such as piping downloads into a shell, reading credential files or hidden Unicode. It is not a guarantee. Review the folder before installing.
What licence does Eval Harness use?
Eval Harness is published under the MIT licence (the repository's licence). It allows redistribution, so the full SKILL.md is shown on this page.
How many tokens does Eval Harness use?
About 974 tokens (SKILL.md is roughly 3.9k characters). Agents keep only the skill's name and description in context until a task matches; then they load SKILL.md in full.
What are the alternatives to Eval Harness?
Skills that share tags, products or a category with Eval Harness: LLM Benchmarking with lm-evaluation-harness (Orchestra-Research/AI-Research-SKILLs, 13k stars), Hugging Face Local Model Evals (huggingface/skills, 11k stars), Looper (ksimback/looper, 710 stars) and Agent Eval Engineering (langchain-ai/langchain-skills, 1.3k stars). The comparison table on this page puts their stars, adoption, token cost, safety result and licence side by side.
Who maintains Eval Harness?
xu-xiang (a GitHub user) maintains it in xu-xiang/everything-claude-code-zh, which has 1,978 GitHub stars. The repository holds 78 skills in this directory. The repository was last updated on March 5, 2026.
Source: xu-xiang/everything-claude-code-zh on GitHub. Facts on this page come from the repository at the commit we read; the author's words are quoted as theirs.