Token导航 LogoToken导航TokenDH.com
研究检索external-servicegithub未标认证来源可访问许可证需确认审计提醒

plugin-engineer插件工程师

Agent Skill

plugin-engineer 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中需要根据关键词、任务场景或来源线索快速定位候选结果时使用。可结合来源仓库、安装命令和原始 README 继续核验具体用法。安装前建议确认权限范围、维护状态,以及是否会触发联网、命令执行或文件读写。

总安装

222

周安装

9

GitHub Stars

26

下载量

70
CodexClaudeCursorGemini CLI

安装说明

本站只整理中文说明和来源信息,不托管安装包,也不代用户安装。

GitHub

来源数

2

许可证

unknown

最后核验

2026-05-01

来源状态

来源可访问

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:plugin-engineer(插件工程师)
来源仓库:https://github.com/outfitter-dev/agents
仓库路径:skills/plugin-engineer
安装命令:
npx skills add https://github.com/outfitter-dev/agents --skill plugin-engineer
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/outfitter-dev/agents --skill plugin-engineer

简介

plugin-engineer 用于查找、检索和筛选相关信息,适合在 Codex、Claude、Cursor、Gemini CLI 中快速定位候选结果。

  • 它适用于插件开发和工程实践场景,可帮助 Agent 根据关键词或任务需求生成结构化输出。
  • 通过自然语言查询或任务描述触发,返回匹配的文档片段或链接供进一步验证。
  • 安装命令为 npx skills add https://github.com/outfitter-dev/agents --skill plugin-engineer,需确认网络访问权限。
  • 使用前应检查数据源可靠性和更新频率,避免依赖过时或不准确的外部信息。

SKILL.md

Plugin Engineer

Transform external repositories into Claude Code plugins.

External Repo → Research → Recon → Patterns → Codify → Author → Package → Audit → Plugin

Steps

  1. Clarify target repo and plugin goals
  2. Load the outfitter:research skill for external discovery (docs, APIs, community patterns)
  3. Load the outfitter:codebase-recon skill for internal analysis of target repo
  4. Load the outfitter:patterns skill to extract repeatable patterns worth automating
  5. Load the outfitter:codify skill to map patterns to component types
  6. Author components using outfitter:skills-dev or outfitter:claude-skills
  7. Load the outfitter:claude-plugins skill to package into distributable plugin
  8. Delegate by loading the outfitter:claude-plugin-audit skill for validation

<when_to_use>

  • Turning a CLI tool into a Claude Code plugin
  • Creating skills that wrap an external library or API
  • Building plugin companions for MCP servers
  • Extracting automation opportunities from a third-party repo
  • Packaging workflow patterns around external tools

NOT for: plugins from scratch (use claude-plugins), single-skill creation (use skills-dev), existing Claude Code plugins (use claude-plugin-audit)

</when_to_use>

Artifact Structure

Track progress with artifacts in artifacts/plugin-engineer/:

artifacts/plugin-engineer/
├── discovery.md      # Research output (docs, APIs, community patterns)
├── recon.md          # Codebase analysis (structure, conventions, key files)
├── patterns.md       # Extracted patterns with automation value
├── mapping.md        # Pattern → component mapping decisions
├── components/       # Authored skills, agents, hooks, commands
│   ├── skill-1/
│   ├── skill-2/
│   └── ...
└── audit.md          # Plugin validation results

Quick Mode

For simple repos (single-purpose CLI, small API wrapper):

  1. Skip stages 3-4 — go direct from recon to authoring
  2. Create 1-2 skills covering primary use cases
  3. Package immediately

Trigger: User says "quick", repo has < 5 main commands, or clear single purpose.

Stages

Load the maintain-tasks skill for stage tracking. Stages advance only.

StageSkillactiveForm
1. Discoveryoutfitter:research"Researching external docs"
2. Reconoutfitter:codebase-recon"Analyzing target repo"
3. Patternsoutfitter:patterns"Extracting patterns"
4. Mappingoutfitter:codify"Mapping to components"
5. Authoringoutfitter:skills-dev"Creating components"
6. Packagingoutfitter:claude-plugins"Packaging plugin"
7. Auditoutfitter:claude-plugin-audit"Validating plugin"

Stage 1: Discovery

Load outfitter:research skill. Gather external docs, community patterns, pain points.

See stage-1-discovery.md for details.

Stage 2: Recon

Load outfitter:codebase-recon skill. Analyze structure, API surface, conventions.

See stage-2-recon.md for details.

Stage 3: Patterns

Load outfitter:patterns skill. Extract workflows, command sequences, decision points.

See stage-3-patterns.md for details.

Stage 4: Mapping

Load outfitter:codify skill. Map patterns to component types (skill, command, hook, agent).

See stage-4-mapping.md for details.

Stage 5: Authoring

Load appropriate skill per component type. Create in artifacts/plugin-engineer/components/.

See stage-5-authoring.md for details.

Stage 6: Packaging

Load outfitter:claude-plugins skill. Create plugin structure with manifest and README.

Ask: "Do you have an existing marketplace to add this plugin to?" If yes, prepare the marketplace entry.

See stage-6-packaging.md for details.

Stage 7: Audit

Delegate by loading outfitter:claude-plugin-audit skill. Validate before distribution.

See stage-7-audit.md for details.

<decision_points>

Key decisions during engineering process:

Which patterns to automate?

  • High frequency + medium complexity = best ROI
  • Low frequency + high complexity = consider if audience is technical
  • One-off patterns = skip

Skills vs Commands?

  • Multi-step, needs guidance → Skill
  • Quick action, obvious usage → Command
  • User entry point to skill → Both (command loads skill)

Include agents?

  • Only for complex repos with orchestration needs
  • Most plugins don't need custom agents
  • Consider if existing agents (analyst, engineer) suffice

Quick mode vs full pipeline?

  • Single-purpose tool → Quick mode
  • Complex tool with many features → Full pipeline
  • Unclear → Start with recon, decide after

</decision_points>

ALWAYS:

  • Start with discovery before touching code
  • Track artifacts at each stage
  • Validate patterns have 3+ use cases before creating components
  • Use existing skills for authoring (don't reinvent)
  • Run audit before declaring complete

NEVER:

  • Skip recon stage (even for familiar repos)
  • Create agents without clear orchestration need
  • Package without audit validation
  • Over-engineer (simpler plugin > feature-complete plugin)

Stage guides:

Skills loaded:

  • outfitter:research — external discovery methodology
  • outfitter:codebase-recon — repo analysis approach
  • outfitter:patterns — pattern extraction
  • outfitter:codify — pattern-to-component mapping
  • outfitter:claude-plugins — plugin packaging

适合场景

01

用户想查找某类 Agent Skill 时

02

需要根据任务场景推荐可安装能力包时

03

需要对比不同来源的安装命令和来源信息时

能力概览

能力 1

按任务关键词查找相关 Skills

能力 2

展示可复制的安装命令

能力 3

保留来源站点、仓库和原始说明,方便继续核验

能力 4

展示第三方安全扫描或审计结果

安装后应在对应宿主中按原始 README 的触发条件使用;具体调用方式请以来源页面和 README 为准。

平台分布

Codex

35.33%
按下载量换算25

Claude

29.08%
按下载量换算20

Cursor

21.41%
按下载量换算15

Gemini CLI

10.47%
按下载量换算7

安全审计

Gen Agent Trust Hub

通过

Socket

通过

Snyk

可疑

权限和风险

external-service

该 Skill 可能调用第三方服务、云服务或外部模型 API,使用前需要确认账号、额度、数据发送范围和服务条款。

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 warning/failed 结果,不能写成本站确认安全。当前只有一个来源,正式发布前建议补源仓库或其他目录站核验。

来源信息

继续浏览同类 Skills