skill-builder
Skill28 starsChanged 3 months agoMITView on GitHub
通过渐进式需求澄清、工作流设计、资源拆分、实现和评测,创建或优化高质量 Agent Skill。适用于“创建一个 skill”“把工作流封装成 skill”“优化 SKILL.md”“补充 scripts、references、assets 或 evals”等请求。Use when a user wants a standards-compliant, maintainable, testable Agent Skill package rather than prompt text alone.
74 other projects use a version of this.
Adopted by
Most recently taken up by
- Kikobazz123/lordgen-skill-buildertheir own version, first seen Nov 8, 2026
- AsiaOstrich/EngramGraphtheir own version, first seen Nov 8, 2026
- shetengteng/skillix-hubtheir own version, first seen Nov 8, 2026
- davidgfolch/AI-job-searchtheir own version, first seen Nov 8, 2026
- RKelln/hermes_skillstheir own version, first seen Nov 7, 2026
- beCyborg/jadlis-skill-buildertheir own version, first seen Nov 6, 2026
Used with
Projects with it also use contract-test, literature-research and 10 more.
Other files people use alongside this one.
Files found in the same public projects as a version of this one, more often than they turn up anywhere else. Each project counts once, large collections count for less, and this file's own repository is left out.
squall-chuaSkillcontract-test
Test a running API against its own contract — OpenAPI, GraphQL, protobuf, or AsyncAPI — from the outside, and report every drift between what the contract promises and what the API does.
Also in 3 projects that use this
nickjwellsSkillliterature-research
Search academic literature and perform deep research reviews. Use when user asks to search PubMed, find academic papers, or do literature reviews.
Also in 3 projects that use this
nopandererSkillcommit-and-push
Automatically analyze git changes, generate Conventional Commits compliant commit messages, commit changes, and push to remote in a single automated workflow. Use when the user invokes /commit-and-push or explicitly requests to automatically commit and push changes following Conventional Commits format (feat:, fix:, docs:, etc.). This skill provides a fully automated git workflow without manual intervention.
Also in 3 projects that use this
MicrosoftSkillskill-reviewer
Review skill PRs with structured severity-rated feedback covering token budgets, routing conflicts, and repo conventions. WHEN: \"review skill\", \"review skill PR\", \"review skill changes\", \"check skill quality\", \"skill PR feedback\".
Also in 3 projects that use this
SkillPilotCHSkilllevel-up
Eine Automatisierung pro Woche finden, scopen, shippen. Output landet als proposed-Eintrag in cadence.md mit konkreten Implementations-Schritten.
Also in 3 projects that use this
MicrosoftSkillexcalidraw-diagram
Create Excalidraw diagram JSON files that make visual arguments. Use when the user wants to visualize workflows, architectures, or concepts.
Also in 4 projects that use this
Other versions74 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 74
- yusufkaraaslan/Skill_Seekers★ 15ksame name
- block/berd★ 1ksame name
- boshu2/agentops★ 453same name
- metaskills/skill-builder★ 112same name
- AsiaOstrich/universal-dev-standards★ 84same name
- Amazon-Quick/Amazon-Quick-official-catalog★ 50same name
- nateherkai/a-bunch-of-skills★ 41same name
- clasen/Skills★ 40same name
- arctan303/creator.skill★ 19same name
- aiagentwithdhruv/skills★ 16same name
64 more not listed
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository4 other files it gives its agents.
The file
Files
Agent Skill 构建器
目标
把一个想法或已有 Skill 直接推进到结构合规、指令清晰、资源克制、脚本可靠、评测可复现的完整能力包。除非关键信息确实无法推断,不要停在建议或模板阶段。
核心原则
- 先读现状:修改已有 Skill 时,先完整读取
SKILL.md、已引用资源、脚本、测试和仓库约定。 - 只问阻塞问题:能从上下文合理推断的内容直接处理;必须提问时每轮不超过 3 个。
- 渐进披露:主文件保留触发、决策、流程和质量门槛,长知识放 references,确定性操作放 scripts。
- 验证比例随风险增长:至少覆盖正常输入、缺失输入和边界输入;共享脚本需要单元测试。
- 描述决定发现:
description同时写清任务、典型触发、边界和交付结果,不塞入完整工作流。 - 自动推进:用户已经授权实施时,各阶段通过后直接进入下一阶段,不重复索要确认。
工作流程
1. 建立任务契约
从用户输入和现有文件中提取:
- 用户要反复完成的具体任务。
- 典型触发话术与相邻但不应触发的请求。
- 必需输入、可选输入和合理默认值。
- 最终交付物、保存位置和完成信号。
- 依赖的工具、网络、凭据、运行时或应用。
- 至少 3 条可判定的质量标准和主要失败模式。
只有触发边界、交付格式或高风险行为仍不明确时才提问。用户给出完整规格时直接进入设计;用户只给一句模糊想法时,优先询问“最终交付什么”“谁会在什么场景触发”“什么算成功”。
产出一份内部任务契约。只有存在会改变实现方向的互斥选择时才展示给用户确认,不把确认仪式当成固定门槛。
2. 设计最小能力包
先区分目标:可移植的 Agent Skills 包只要求根目录有 SKILL.md;下面是需要适配 Codex 展示并满足本仓库发布门禁时采用的 Sanqi 完整结构,不是开放标准的强制目录:
skill-name/
├── SKILL.md
├── agents/
│ └── openai.yaml # 仅在目标客户端需要时
├── scripts/ # 仅在确定性操作能提高可靠性时添加
├── references/ # 仅放按需读取的领域知识或长规范
├── assets/ # 仅放最终输出会直接使用的模板或素材
└── evals/ # 本仓库的发布约定
├── evals.json
└── trigger-evals.json不要为了目录完整而创建空目录。脚本、引用和资产必须在主文件中说明何时使用;每个引用从 SKILL.md 直接链接,避免多层引用链。
选择适合的工作流模式时读取 references/design-patterns.md。
3. 编写标准元数据
顶层使用 Agent Skills 支持的字段。最小可移植包需要 name 和 description;license、compatibility、metadata 属于可选字段。本仓库另外要求 MIT 和统一的展示元数据,不能把这些仓库约定称为开放标准的强制要求:
---
name: skill-name
description: >
[做什么;用户何时会需要;典型话术;相邻边界;最终交付什么。]
license: MIT
compatibility: [运行时、工具、网络或应用要求。]
metadata:
author: "author-name"
version: "1.0.0"
---要求:
name与目录名完全一致,只含小写字母、数字和单连字符,长度不超过 64。description非空且不超过 1024 字符,使用具体名词和用户话术,不依赖只有作者懂的简称。- 自定义展示信息放进
metadata,所有值使用字符串。 - 仅在 Skill 确实需要时声明
allowed-tools;不要用它掩盖未说明的依赖。
目标是 Codex 等使用该文件的客户端,或发布到本仓库时,补充 agents/openai.yaml。本仓库要求其中字符串加引号,default_prompt 明确提到 $skill-name;其他客户端可能忽略此文件。
4. 编写可执行指令
正文使用祈使语气,包含:
- 一句话目标。
- 开始条件和输入缺失策略。
- 有顺序、分支和产出物的工作流。
- 错误、降级、重试和停止条件。
- 可用“通过/不通过”判断的质量标准。
- 最终交付内容、文件路径和已知局限。
避免重复常识、空泛角色扮演和无法验证的“高质量”“深入分析”。正文接近 500 行时必须拆分;引用文件只保留完成任务所需的信息。
5. 实现确定性资源
以下内容优先写成脚本:解析、转换、命名、批处理、结构验证、统计和可重复渲染。脚本必须:
- 使用明确参数和非零失败码。
- 不依赖作者机器上的固定绝对路径。
- 对覆盖文件、网络失败、无效输入和缺失依赖给出可执行错误信息。
- 在至少一个正常案例和一个失败案例上实际运行。
领域知识、API 规范、长模板和评分量表放 references。输出模板或二进制素材放 assets。删除未被工作流使用的资源。
6. 建立评测
发布到本仓库的每个 Skill 至少提供以下数据集;其他仓库按自身评测政策调整:
- 触发评测:不少于 6 个正例和 6 个难负例,覆盖中英文、口语化请求和相邻能力混淆。
- 输出评测:不少于 3 个案例,包含标准场景、缺口场景和失败/边界场景。
- 每个输出案例包含可观察 assertions,不以“看起来不错”作为标准。
条件允许时,用仓库评测脚本进行元数据路由模拟;拿到真实执行记录后再运行输出 assertions 评测。路由模拟不等同于目标客户端的真实 Skill 自动触发,输出评分也不等同于端到端执行。对代表性案例可做启用 Skill 与不启用 Skill 的对照运行,记录成功率、遗漏和副作用。没有模型或凭据时仍要校验评测 JSON 的结构,并明确说明尚未运行模型评测。
7. 验证并回修
按仓库已有工具优先执行:
- 官方或仓库级 Skill 结构校验。
- 所有脚本的语法检查和单元测试。
- 所有本地引用路径存在性检查。
- 评测数据 schema 检查。
- README、目录或清单同步检查。
- Git diff 审查,确认没有缓存、密钥、用户产物和无关改动。
任一确定性门禁失败时直接回修并重跑。外部服务、凭据或视觉人工判断无法自动验证时,列为残余风险,不伪造通过结果。
质量标准
- 触发描述能区分至少一个相邻但不应触发的请求。
- 顶层元数据符合开放标准;本仓库额外约定单独通过仓库验证器检查。
- 主流程、降级路径、失败条件和最终交付都明确。
- 所有引用存在,脚本可独立执行且有失败码。
- 至少 12 个触发案例和 3 个输出案例通过结构校验。
- 确定性测试与仓库发布门禁全部通过。
- 最终报告区分“已验证”“需外部环境验证”和“已知局限”。
最终反馈
向用户报告 Skill 名称、实际改动、关键设计决策、测试命令与结果、未运行的模型/外部集成评测,以及生成文件的完整路径。不要重复整份任务契约,也不要在任务已经完成后强制追加确认步骤。
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

