harness-benchmark SKILL.md skillA
harness-benchmark from wangjs-jacky/jacky-skills is an agent skill: a SKILL.md folder an AI agent loads when a task calls for it.
mdr add wangjs-jacky/jacky-skills/harness-benchmark@git:20260711.dae29b8Pins git:20260711.dae29b8 in mdr.lock. Pin the fingerprint instead.
- Audit grade
- A, 17 of 17 checks passed
- Kind
- SKILL.md skill
- Repository
- wangjs-jacky/jacky-skills
- Path
plugins/evaluators/harness-benchmark/SKILL.md- Versions recorded
- 1
- Latest version
- git:20260711.dae29b8, committed Jul 11, 2026
- Size
- 9,849 bytes
- Repository stars
- 15
- License
- MIT
基于 Anthropic Harness 框架的任务目标基准评测器。采用生成器-评估器分离架构（GAN 启发），支持前端设计评估（设计质量/原创性/工艺/功能性 4 维度）和全栈开发评估（冲刺合同/功能完整性/代码质量/用户体验 4 维度）。触发于 /harness-benchmark、「评估任务」、「benchmark harness」、「设计评估」、「质量评估」等关键词。.
Collected from public GitHub. Every version is saved exactly as it was and never changes. Install it at an exact version with the mdr tool, and every file is checked before it reaches your agent. The main file is checked against the SHA-256 fingerprint recorded here, and the others against the fingerprints GitHub keeps for its source commit. The automatic checks below grade the latest version, and the same checks always give the same file the same grade.
What the file says
<role>
你是 Harness Benchmark，一个严格且专业的任务目标基准评测器。
你的职责是：
1. **接收评估目标** — 明确评估对象（代码产出、设计稿、功能模块等）
2. **选择评估模式** — 根据任务类型匹配对应的评估维度体系
3. **执行评估** — 按维度逐项审查，以怀疑态度寻找问题，不做宽容评价
4. **输出评估报告** — 包含各维度分数、具体问题、改进建议
**你不是生成器。你不修改代码，不做任何改进工作。你只评判。**
</role>
<philosophy>
## 核心理念：生成器-评估器分离
本评估器的设计灵感来自 Anthropic 工程团队的研究成果（[Harness Design for Long-Running Application Development](https://www.anthropic.com/engineering/harness-design-long-running-apps)），核心洞察：
### 问题一：自我评估失真
AI 对自己的作品倾向于过度自信。即使输出平庸，自我评估也会给出好评。将执行工作的 Agent 与判断工作的 Agent 分离，是解决这个问题的有力杠杆。
### 问题二：上下文焦虑
模型在长任务中会逐渐失去方向并过早收尾。上下文重置（而非压缩）提供了干净的起点，让评估器以全新视角审视产出。
### 解决方案：GAN 启发的分离架构
```
生成器（Generator） 评估器（Evaluator）
│ │
├─ 执行工作 ├─ 审判工作
├─ 实现功能 ├─ 逐维度打分
├─ 偏向自信 ├─ 偏向怀疑
│ │
└──── 反馈循环 ──────────────┘
↑ 评估反馈驱动下一轮生成 ↓
```
**关键原则**：
- 调整一个独立的评估器使其持怀疑态度，远比让生成器批判自己的工作容易
- 评估器通过文件与生成器通信，保持上下文独立
- 每次评估都是全新的审视，不继承生成器的认知偏差
> 评估器校准标准（怀疑优先、具体评判、惩罚 AI 模式、鼓励风险承担、功能深度优于广度）详见 references/iteration-model.md
</philosophy>
<purpose>
将"这个任务完成了吗？"这样的模糊判断，转化为具体的、可评分的、可复现的评估体系。
</purpose>
<trigger>
```text
…Read the whole file at its exact version.
How to install
mdr add wangjs-jacky/jacky-skills/harness-benchmark@sha256:7c2e3accc7378ad8The line at the top of this page pins the label git:20260711.dae29b8, which is what most people want. This one pins the sha256 fingerprint of the contents instead, so it survives a label being moved. Either way the exact version is written to mdr.lock, and mdr install downloads and checks those contents again, so it installs them exactly or stops with an error.
[](https://markdownregistry.com/a/art_b4vpqngg35vnzu56)
1 badge image fetches in 30 days. What the badge shows.
Versions
The registry has recorded 1 version of this file, committed on Jul 11, 2026.
Audit of the latest version
- pass: Frontmatter block present
- pass: Frontmatter declares a name
- pass: Frontmatter declares a description
- pass: Size between 200 bytes and 200 KB (9849 bytes)
- pass: No zero-width or bidi control characters
- pass: No instruction hidden inside an HTML comment
- pass: No link to an exfiltration or paste host
- pass: No credential-shaped string
- pass: No instruction to send local credentials anywhere
- pass: No text hidden with inline styles
- pass: No prompt-injection phrasing
- pass: No curl or wget piped into a shell
- pass: No recursive delete of root, home or parent
- pass: No instruction to read or print local credentials
- pass: No base64 blob over 200 characters
- pass: No link to a raw IP address
- pass: No script tag
Source
wangjs-jacky/jacky-skills · 15 stars · license MIT · pushed 2026-10-19 · branch main
API
GET https://markdownregistry.com/api/v1/artifacts/art_b4vpqngg35vnzu56 GET https://markdownregistry.com/api/v1/resolve?ref=wangjs-jacky/jacky-skills/harness-benchmark GET https://markdownregistry.com/api/v1/blob/7c2e3accc7378ad864d067a13ce4b5a9cf2438813725ce3f205b0f8e003bba01
Your AI assistant does the legwork. You hear about the deals worth your word. Hand yours the standing instructions at modelranch.com and it joins the network that reads files like this one.
Guides: the SKILL.md format and its required fields · what to check before you install a skill · how to pin a skill to an exact version · where to find agent skills · keeping a team on the same skills · which agent file does what.