Token导航 LogoToken导航

Agent Skill · 研究检索

api-workshop(API workshop 搜索)

用于辅助 API 设计、接口文档、请求响应结构和服务集成说明。它适合让 Agent 梳理 endpoint、生成 OpenAPI 草稿、检查字段命名、整理错误码或辅助前后端联调。使用时需要确认真实业务语义、鉴权方式、分页和错误处理规则;涉及生成接口文档时,应避免凭空补字段,最好从现有代码、schema 或接口样例中提取事实。

api-workshop 是研究检索类 Agent Skill,可用宿主为 Codex、Claude、Cursor、Gemini CLI,通过GitHub 仓库获取,权限为执行命令;许可证信息还不明确,第三方扫描有提醒项,按安装量在研究检索类的 73,675 个 Skill 里排第 66,577。

总安装 261周安装 11GitHub Stars 191下载 92核验 2026-05-01

安装说明

本站只整理中文说明,不托管安装包,也不代用户安装。
GitHub
许可证
unknown

安装方式

通过对话安装

复制提示词发给支持本地命令或 Skills 的 AI 助手,先确认命令和权限,再让它执行。

请帮我安装这个 Agent Skill:api-workshop(API workshop 搜索)
来源仓库:https://github.com/wcygan/dotfiles
仓库路径:skills/api-workshop
安装命令:
npx skills add https://github.com/wcygan/dotfiles --skill api-workshop
安装前请先检查当前环境是否支持对应 CLI,并向我确认将要执行的命令、安装目录、联网范围和文件读写权限;确认后再执行。

命令行安装

复制命令到本机终端执行。该命令会通过 npx skills 从第三方来源获取 Skill;本站只展示命令,不托管安装包,也不自动执行。

skills.shnpx skills
npx skills add https://github.com/wcygan/dotfiles --skill api-workshop

这个 Skill 是做什么的

研究检索类 Skill 主要解决「找资料、核对事实、把来源整理清楚」的问题,通常会让助手先检索再总结。

组织多 Agent 参与的 API 设计与评审工作坊。通过辩论机制产出经过充分讨论的 API 规范。内置 OpenAPI 模板与争议解决提示词。适用于复杂系统集成前的架构决策阶段。api-workshop 属于研究检索类 Skill,可作为该场景下的辅助能力补充。需要至少五个专用角色 Agent 协同运行。

来源仓库内路径:skills/api-workshop

上手步骤

4 步
  1. 1

    确认自己的宿主工具

    公开资料标注支持 Codex、Claude、Cursor、Gemini CLI;先确认自己用的是其中之一,再按对应工具的 Skills 目录放置。

  2. 2

    拿到安装命令

    可以直接复制的命令是 npx skills add https://github.com/wcygan/dotfiles --skill api-workshop;命令里的占位符换成自己的路径或密钥再执行。

  3. 3

    安装后重启宿主

    命令执行完要完全退出宿主工具再打开,让新装的 Skill 被重新扫描到,否则列表里可能看不到。

  4. 4

    按最小权限试用

    它需要执行命令,先在测试目录或测试账号里跑,确认没问题再放到正式项目。

常见问题

6 条

api-workshop 怎么安装?

在终端执行 npx skills add https://github.com/wcygan/dotfiles --skill api-workshop。装完完全退出宿主工具再打开,让它重新扫描本地 Skill。

api-workshop 支持哪些 AI 工具?

公开资料标注支持 Codex、Claude、Cursor、Gemini CLI。其他工具只要能读取本地 Skill 目录,一般也能手动放进去,但要自己确认触发方式是否一致。

api-workshop 需要什么权限?

权限级别是执行命令。安装流程涉及命令执行,可能通过 npx skills add https://github.com/wcygan/dotfiles --skill api-workshop 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

api-workshop 是免费的吗?

公开资料没有标注明确许可证,收费和商用限制都无法确认。正式使用前请到来源仓库核对授权说明。

api-workshop 用的人多吗?

公开数据是累计安装 261 次,近一周安装 11 次,下载 92 次,GitHub 191 星。这些数字来自第三方目录统计,只能当作热度参考,不代表质量或安全性。

api-workshop 安全吗?

第三方扫描有提醒项,说明存在需要人工确认的行为。安装前请打开源码看清它读了什么、联网到哪里。

值得选的地方和要先想清楚的地方

值得选的地方

  • 同时适配 4 种工具,换客户端不用重新找替代品。

先想清楚的地方

  • 需要执行命令,装到常用环境前要先评估影响面。
  • 许可证没有明确标注,能不能商用、能不能再分发都不确定。
  • 第三方扫描存在提醒项或未通过项,需要人工逐条确认。
  • 第三方目录整理,版本更新和行为细节以来源仓库为准。

适不适合你

更适合

  • 已经在用 Codex,想直接扩展它能力的人。
  • 需要处理研究检索类任务、又不想每次都手写提示词的人。

不太适合

  • 不能接受 Skill 在你机器上执行命令的人。
  • 需要明确授权才能用到公司项目里的场景。
  • 需要厂商级技术支持和 SLA 的生产系统。

研究检索类 Skill 对比

6 个同类
Skill总安装GitHub 星权限许可证扫描
find-skills

寻找技能 · Codex、Claude

31,199,81316,486操作浏览器MIT有提醒
self-improving-agent

自我改进 Agent · OpenClaw

16,464,0023,396接触敏感数据MIT-0通过
soultrace

灵魂追踪 · Codex、Claude

7,588,94412需要联网未明确有提醒
self-improving

自我完善 · OpenClaw

6,829,2841,046需要联网MIT-0通过
microsoft-foundry

微软代工厂 · Codex、Claude

6,614,144764调用外部服务未明确有提醒
api-workshop

API workshop 搜索 · 本页

261191执行命令未明确有提醒

按安装量取同分类前几名;数字来自第三方目录统计,只作热度参考,不代表质量结论。

SKILL.md 原文

API Workshop

A debate-driven multi-agent workshop for designing and reviewing APIs. Five specialized agents propose designs and challenge each other to produce a well-reasoned API specification.

Prerequisites

Agents will need access to API design patterns and standards. The REFERENCE.md file provides OpenAPI templates, design principles, and debate prompts.

Inputs

The user provides:

  1. API Description (required) — What the API should do, or URL to existing spec
  2. Context (optional) — Business domain, constraints, existing systems
  3. Focus areas (optional) — Which aspects matter most (see Agent Roster below)
  4. Format preference (optional) — OpenAPI 3.x, GraphQL schema, AsyncAPI

If the user provides a URL to an existing spec, agents will review it rather than design from scratch.

Agent Roster

Each agent brings a specialized perspective and debates with others:

#AgentFocusChallenges
1API DesignerConsumer UX, discoverability, naming"Is this intuitive for developers?"
2Domain ModelerCorrect modeling, consistency, DDD"Does this match the domain?"
3Security AuditorAuth/authz, data exposure, OWASP"What are the security risks?"
4Performance AnalystEfficiency, caching, over/under-fetching"Will this scale?"
5Tech LeadStandards compliance, versioning, docs"Does this follow best practices?"

Workflow

Phase 1: Context Gathering (Main Agent)

Before spawning the workshop agents:

  1. If given a description: Extract requirements, constraints, and success criteria
  2. If given a URL: Fetch the existing API spec and summarize it
  3. Identify domain entities: List the core resources/types involved
  4. Clarify open questions: Ask user if anything is ambiguous
  5. Create initial context document: workspace/api-context.md

Phase 2: Initial Proposals (Sub-agents 1-5)

Spawn all five agents in parallel. Each receives:

  • The context document from Phase 1
  • Their agent instructions (from agents/*.md)
  • The API design reference (REFERENCE.md)

Each agent independently:

  1. Proposes an initial design approach
  2. Creates example endpoint/schema definitions
  3. Writes their rationale and concerns
  4. Saves to workspace/proposals/<agent-name>.md

Phase 3: Cross-Examination (Main Agent Orchestrates)

For each significant design decision, run a mini-debate:

  1. API Style (REST vs GraphQL vs RPC)

- API Designer proposes based on UX - Performance Analyst challenges based on query patterns - Tech Lead weighs standardization concerns

  1. Authentication Strategy (OAuth2, API keys, JWT)

- Security Auditor proposes approach - API Designer challenges if it adds friction - Tech Lead validates against standards

  1. Versioning Strategy (URI, header, content negotiation)

- Tech Lead proposes approach - API Designer challenges impact on consumers - Domain Modeler validates semantic compatibility

  1. Resource Design (granularity, relationships, operations)

- Domain Modeler proposes structure - Performance Analyst challenges N+1 issues - API Designer validates developer experience

  1. Error Handling (status codes, error formats, problem details)

- Security Auditor proposes secure defaults - API Designer ensures clarity for consumers - Tech Lead validates RFC 7807 compliance

For each debate:

  • Present the question and stakes
  • Let 2-3 agents argue their positions
  • Synthesize the decision with rationale
  • Document trade-offs considered

Phase 4: Synthesis (Main Agent)

After all debates complete:

  1. Reconcile proposals: Merge agent input into coherent design
  2. Resolve conflicts: Make final calls on contested decisions
  3. Document rationale: Explain why each choice was made
  4. Identify risks: Note what could go wrong and mitigations
  5. Generate specification: Create complete API spec in requested format

Phase 5: Output

Generate the final deliverable following the template in REFERENCE.md:

  1. Executive Summary: High-level design decisions and rationale
  2. Architecture: API style, auth, versioning, error handling
  3. Specification: Complete OpenAPI/GraphQL schema
  4. Design Rationale: Why each major decision was made
  5. Trade-offs: What was considered and rejected
  6. Implementation Notes: Tips for backend developers
  7. Client Examples: Sample requests/responses in multiple languages

Save to workspace/api-specification.md and present to the user.

Debate Topics

The workshop should address these key decisions:

API Style

  • REST: Resource-oriented, HTTP semantics, wide tooling
  • GraphQL: Flexible queries, type-safe, learning curve
  • RPC: Simple, efficient, less discoverable

Versioning

  • URI versioning: /v1/users (explicit, breaks caching)
  • Header versioning: Accept: application/vnd.api+json;version=1 (clean URIs, harder to test)
  • Content negotiation: Media types control breaking changes

Authentication

  • OAuth2: Standard, complex setup, great for delegated auth
  • API keys: Simple, less secure, good for server-to-server
  • JWT: Stateless, revocation challenges, flexible claims

Pagination

  • Offset/limit: Simple, inconsistent under writes
  • Cursor-based: Consistent, opaque, harder to jump to page N
  • Keyset: Efficient, requires indexed sort key

Error Handling

  • Problem Details (RFC 7807): Structured, extensible, standard
  • Custom format: Tailored to domain, non-standard
  • HTTP status only: Simple, lacks detail

Output Structure

workspace/
├── api-context.md                  # Phase 1 context gathering
├── proposals/
│   ├── api-designer.md
│   ├── domain-modeler.md
│   ├── security-auditor.md
│   ├── performance-analyst.md
│   └── tech-lead.md
├── debates/
│   ├── api-style-debate.md
│   ├── auth-strategy-debate.md
│   ├── versioning-debate.md
│   ├── resource-design-debate.md
│   └── error-handling-debate.md
└── api-specification.md            # Final deliverable

Coordinator Responsibilities

  1. Gather context and clarify requirements with user
  2. Spawn proposal agents in parallel
  3. Orchestrate debates on key design decisions
  4. Ensure all agents engage (not just API Designer)
  5. Synthesize final specification with rationale
  6. Present deliverable to user
  7. Offer to generate client SDKs or mock server if requested

Customization

The user can customize the workshop by:

  • Skipping agents: "Skip performance, focus on security"
  • Adding constraints: "Must support GraphQL subscriptions"
  • Specifying style: "Design a REST API following JSON:API spec"
  • Setting priorities: "Optimize for developer experience over efficiency"
  • Providing examples: "Similar to Stripe's API" or "Here's our existing API"
  • Requesting specific debates: "Debate whether we need versioning at all"

Adapt the agent roster and debate topics accordingly.

Example Invocations

/api-workshop design an API for a task management system

/api-workshop review https://api.example.com/openapi.json

/api-workshop design a GraphQL API for e-commerce with subscriptions

/api-workshop compare REST vs GraphQL for our social network

The workshop adapts to whether you're designing new, reviewing existing, or comparing alternatives.

平台分布

Codex

35.75%
按下载量换算33

Claude

28.86%
按下载量换算27

Cursor

19.49%
按下载量换算18

Gemini CLI

9.27%
按下载量换算9

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

执行命令

安装流程涉及命令执行,可能通过 npx skills add https://github.com/wcygan/dotfiles --skill api-workshop 联网下载 Skill 或依赖。用户安装前应确认命令来源、仓库内容和执行环境。

安装前确认

本站仅展示第三方公开信息,不托管安装包,不提供自动安装或运行环境。安装前应自行审查源码、依赖和命令行为。来源安全扫描存在提醒或未通过项,安装前请自行复核。目前只有一个公开来源,建议同时核对源仓库。

继续浏览同类 Skills