Token导航 LogoToken导航TokenDH.com
研究检索敏感数据github未标认证来源可访问clear审计未展示

claude-plugin-developmentClaude plugin 开发

Agent Skill

claude-plugin-development 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

544

周安装

22

GitHub Stars

公开资料未说明

下载量

171
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

MIT

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:claude-plugin-development(Claude plugin 开发)
来源仓库:https://github.com/outfitter-dev/agents
仓库路径:skills/claude-plugin-development
安装命令:
npx skills add outfitter-dev/agents --skill "claude-plugin-development"
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。

AgentSkills.tonpx skills
npx skills add outfitter-dev/agents --skill "claude-plugin-development"

简介

claude-plugin-development 提供插件开发相关的信息检索服务。

  • 支持根据开发场景或技术栈匹配相关文档与最佳实践。
  • 使用 npx skills add outfitter-dev/agents --skill "claude-plugin-development" 安装。
  • 需关注是否包含代码生成、文件修改或远程依赖拉取等操作。
  • 建议先查阅源码了解其知识库覆盖范围与更新策略。

SKILL.md

Claude Plugin Development

Complete lifecycle for developing, validating, and distributing Claude Code plugins.

Quick Start

# 1. Scaffold plugin
./scripts/scaffold-plugin.sh my-plugin --with-commands

# 2. Add components (commands, agents, hooks, skills)
# 3. Test locally
/plugin marketplace add ./my-plugin
/plugin install my-plugin@my-plugin

# 4. Distribute
git push origin main --tags

Lifecycle Overview

Discovery -> Init -> Components -> Validate -> Distribute -> Marketplace
    |         |          |            |            |             |
    v         v          v            v            v             v
 Purpose   Scaffold   Commands    Structure    Package      Catalog
  Scope    plugin.json  Agents     Testing     Version      Publish
  Type      README      Hooks      Quality     Release       Share

Phase 1: Discovery

Before creating a plugin, clarify:

QuestionImpact
What problem does this solve?Plugin scope and features
Who will use it?Distribution method
What components are needed?Commands, agents, hooks, MCP servers
Where will it live?Personal, project, or marketplace

Phase 2: Initialization

Minimal Plugin Structure

my-plugin/
├── plugin.json          # Required: metadata
├── README.md            # Required for distribution
└── commands/            # Optional components

plugin.json (Required)

{
  "name": "my-plugin",
  "version": "1.0.0",
  "description": "Brief description of what this plugin does",
  "author": {
    "name": "Your Name",
    "email": "you@example.com"
  },
  "license": "MIT"
}

Using the Scaffold Script

./scripts/scaffold-plugin.sh my-plugin \
  --author "Your Name" \
  --with-commands \
  --with-agent \
  --with-hooks

See references/structure.md for complete plugin structure and plugin.json schema.

Phase 3: Components

Add components based on plugin needs. For detailed component authoring, load the appropriate skill:

Slash Commands

Create custom commands in commands/ directory.

Example: commands/review.md

---
description: "Review code for quality issues"
---

Review the following code: {{0}}

Check for:
- Code style and formatting
- Potential bugs
- Performance issues
- Security concerns

For complex commands, load the claude-command-authoring skill.

Custom Agents

Define specialized agents in agents/ directory.

Example: agents/security-reviewer.md

---
name: security-reviewer
description: "Security-focused code reviewer"
---

You are a security expert. When reviewing code:
1. Check for vulnerabilities
2. Verify input validation
3. Review authentication flows
4. Report issues with severity levels

For agent design patterns, load the claude-agent-development skill.

Event Hooks

React to Claude Code events via plugin.json:

{
  "hooks": {
    "PreToolUse": [
      {
        "matcher": "Write|Edit",
        "hooks": [
          {
            "type": "command",
            "command": "${CLAUDE_PLUGIN_ROOT}/hooks/validate.sh"
          }
        ]
      }
    ]
  }
}

Hook types: PreToolUse, PostToolUse, PrePromptSubmit, PostPromptSubmit

For hook implementation, load the claude-hook-authoring skill.

Skills

Add reusable methodology patterns in skills/ directory.

For skill authoring, load the skills-development skill.

MCP Servers

Integrate MCP servers for external capabilities:

{
  "mcpServers": {
    "my-server": {
      "command": "${CLAUDE_PLUGIN_ROOT}/servers/my-server",
      "args": ["--config", "${CLAUDE_PLUGIN_ROOT}/config.json"],
      "env": {
        "API_KEY": "${MY_API_KEY}"
      }
    }
  }
}

Path variables:

  • ${CLAUDE_PLUGIN_ROOT} - Plugin installation directory
  • ${MY_API_KEY} - Environment variable expansion

Phase 4: Validation

Before distribution, validate the plugin:

Validation Checklist

Structure:

  • plugin.json exists and is valid JSON
  • Required fields present (name, version, description)
  • Plugin name matches directory name (kebab-case)

Commands:

  • YAML frontmatter with description
  • Parameter syntax correct ({{0}}, {{1}})

Agents:

  • YAML frontmatter with name and description
  • Tool restrictions appropriate

Hooks:

  • Scripts executable (chmod +x)
  • JSON input/output format correct
  • Matchers are valid regex

Documentation:

  • README.md with installation instructions
  • CHANGELOG.md for version history
  • LICENSE file included

Validation Commands

# Validate JSON
jq empty plugin.json

# Check command frontmatter
for f in commands/**/*.md; do
  head -n 5 "$f" | grep -q '^---$' || echo "Missing: $f"
done

# Verify scripts executable
for f in hooks/**/*.sh; do
  test -x "$f" || echo "Not executable: $f"
done

Local Testing

# Add as local marketplace
/plugin marketplace add ./my-plugin

# Install and test
/plugin install my-plugin@my-plugin

# Test commands
/my-command arg1 arg2

Phase 5: Distribution

Semantic Versioning

Follow semver (MAJOR.MINOR.PATCH):

  • MAJOR: Breaking changes
  • MINOR: New features (backward compatible)
  • PATCH: Bug fixes

Release Workflow

# 1. Update version in plugin.json
# 2. Update CHANGELOG.md
# 3. Commit
git add plugin.json CHANGELOG.md
git commit -m "chore: release v1.0.0"

# 4. Tag
git tag v1.0.0

# 5. Push
git push origin main --tags

# 6. Create GitHub release
gh release create v1.0.0 \
  --title "v1.0.0" \
  --notes "Initial release"

Distribution Methods

MethodBest ForSetup
GitHub repoPublic/team pluginsPush to GitHub
Git URLGitLab, BitbucketFull URL in source
Local pathDevelopment/testingRelative path
ZIP packageOffline distributionCreate archive

See references/distribution.md for packaging, CI/CD, and release automation.

Phase 6: Marketplace

What is a Marketplace?

A catalog of plugins defined in .claude-plugin/marketplace.json that enables discovery, installation, and version management.

Creating a Marketplace

mkdir -p .claude-plugin

.claude-plugin/marketplace.json:

{
  "name": "my-marketplace",
  "owner": {
    "name": "Team Name",
    "email": "team@example.com"
  },
  "plugins": [
    {
      "name": "my-plugin",
      "source": "./plugins/my-plugin",
      "description": "Plugin description",
      "version": "1.0.0"
    }
  ]
}

Plugin Sources

Relative path:

{"source": "./plugins/my-plugin"}

GitHub:

{
  "source": {
    "source": "github",
    "repo": "owner/plugin-repo",
    "ref": "v1.0.0"
  }
}

Git URL:

{
  "source": {
    "source": "url",
    "url": "https://gitlab.com/team/plugin.git"
  }
}

Team Configuration

Configure automatic marketplace installation in .claude/settings.json:

{
  "extraKnownMarketplaces": {
    "team-tools": {
      "source": {
        "source": "github",
        "repo": "company/claude-plugins"
      }
    }
  }
}

Marketplace Commands

# Add marketplace
/plugin marketplace add owner/repo
/plugin marketplace add ./local-path

# List available
/plugin marketplace list

# Install from marketplace
/plugin install plugin-name@marketplace-name

# Update
/plugin marketplace update marketplace-name

See references/marketplace.md for full schema, team setup, and hosting strategies.

Best Practices

Naming Conventions

  • Plugin name: kebab-case (e.g., dev-tools)
  • Commands: kebab-case (e.g., review-pr)
  • Agents: kebab-case (e.g., security-reviewer)
  • Scripts: kebab-case with extension (e.g., validate.sh)

Security

  • Never hardcode secrets in plugin files
  • Use environment variables for sensitive data
  • Validate all user inputs in hooks
  • Review third-party dependencies
  • Document security requirements

Documentation

  • README.md: Overview, installation, usage examples
  • CHANGELOG.md: Version history with semver
  • LICENSE: Appropriate license file
  • Commands/Agents: Clear description in frontmatter

Testing

  • Test all commands with various inputs
  • Verify hooks don't block normal workflow
  • Check MCP servers connect properly
  • Test installation and removal
  • Validate cross-platform compatibility

Troubleshooting

Plugin not loading:

  • Verify plugin.json syntax: jq empty plugin.json
  • Check plugin name matches directory
  • Ensure required fields present

Commands not appearing:

  • Verify YAML frontmatter exists
  • Check files in commands/ directory
  • Ensure markdown syntax correct

Hooks not executing:

  • Check scripts executable: chmod +x
  • Verify matcher regex correct
  • Test hook script independently
  • Review JSON output format

MCP servers failing:

  • Verify server binary exists
  • Check environment variables set
  • Test with MCP Inspector
  • Review logs: ~/Library/Logs/Claude/

References

Related Skills

  • claude-command-authoring - Slash command development
  • claude-agent-development - Custom agent design
  • claude-hook-authoring - Event hook implementation
  • skills-development - Skill creation patterns

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

04

需要参考平台分布和安装热度时

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

补充不同宿主或平台的使用分布数据

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

Claude Code

32.15%
按下载量换算55

windsurf

22.2%
按下载量换算38

trae

17.32%
按下载量换算30

OpenCode

13.24%
按下载量换算23

Cursor

7.67%
按下载量换算13

Codex

3.23%
按下载量换算6

安全审计

暂无安全审计结果可展示。

权限和风险

敏感数据

该 Skill 可能接触密钥、Token、环境变量或敏感配置,应进入高风险复核队列,默认不自动发布。

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。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