Files
claude-howto/07-plugins
Luong NGUYEN 6d1e0ae4af refactor(ci): shift quality checks to pre-commit, CI as 2nd pass (#34)
* refactor(ci): shift quality checks to pre-commit, CI as 2nd pass

- Remove ci.yml (lint, security, pytest were only for EPUB scripts)
- Move EPUB build to pre-commit local hook (runs on .md changes)
- Add check_cross_references.py, check_mermaid.py, check_links.py scripts
- Add markdown-lint, cross-references, mermaid-syntax, link-check as
  pre-commit hooks — mirrors all 4 CI doc-check jobs locally
- Remove spell check job from docs-check.yml (breaks on translations)
- Refactor docs-check.yml to reuse scripts/ instead of inline Python
- Add .markdownlint.json config shared by pre-commit and CI
- Update CONTRIBUTING.md with required dependencies and hook table

* fix(ci): resolve all CI check failures in docs-check workflow

- fix(check_cross_references): skip code blocks and inline code spans
  to avoid false positives from documentation examples; fix emoji
  heading anchor generation (rstrip not strip); add blog-posts,
  openspec, prompts, .agents to IGNORE_DIRS; ignore README.backup.md
- fix(check_links): strip trailing Markdown punctuation from captured
  URLs; add wikipedia, api.github.com to SKIP_DOMAINS; add placeholder
  URL patterns to SKIP_URL_PATTERNS; add .agents/.claude to IGNORE_DIRS
- fix(check_mermaid): add --no-sandbox puppeteer config support via
  MERMAID_PUPPETEER_NO_SANDBOX env var for GitHub Actions Linux runners
- fix(docs-check.yml): pass MERMAID_PUPPETEER_NO_SANDBOX=true to mermaid job
- fix(content): repair broken anchors in README.md, 09-advanced-features;
  fix #plugins -> #claude-code-plugins in claude_concepts_guide.md;
  remove non-existent ./docs/performance.md placeholder links; fix
  dependabot alerts URL in SECURITY_REPORTING.md; update auto-mode URL
  in resources.md; use placeholder pattern for 07-plugins example URL
- remove README.backup.md (stale file)

* fix(check-scripts): fix strip_code_blocks regex and URL fragment handling

- fix regex in strip_code_blocks to avoid conflicting MULTILINE+DOTALL
  flags that could fail to strip indented code fences; use DOTALL only
- strip URL fragments (#section) before dispatching link checks to avoid
  false-positive 404s on valid URLs with anchor fragments

* fix(check-scripts): fix anchor stripping, cross-ref enforcement, and mermaid temp file cleanup

- heading_to_anchor: use .strip("-") instead of .rstrip("-") to also strip leading hyphens
  produced by emoji-prefixed headings, preventing false-positive anchor errors
- check_cross_references: always exit with main()'s return code — filesystem checks
  should block pre-commit unconditionally, not silently pass on errors
- check_mermaid: wrap file-processing loop in try/finally so the puppeteer config
  temp file is cleaned up even if an unexpected exception (e.g. UnicodeDecodeError) occurs
- docs-check.yml: remove now-unused CROSS_REF_STRICT env var

* fix(scripts): fix anchor stripping and mermaid output path

- Replace .strip('-') with .rstrip('-') in heading_to_anchor() so leading
  hyphens from emoji-prefixed headings are preserved, matching GitHub's
  anchor generation behaviour.
- Use Path.with_suffix('.svg') in check_mermaid.py instead of
  str.replace('.mmd', '.svg') to avoid replacing all occurrences of .mmd
  in the full temp path.
2026-04-02 02:20:45 +02:00
..

Claude How To

Claude Code Plugins

This folder contains complete plugin examples that bundle multiple Claude Code features into cohesive, installable packages.

Overview

Claude Code Plugins are bundled collections of customizations (slash commands, subagents, MCP servers, and hooks) that install with a single command. They represent the highest-level extension mechanism—combining multiple features into cohesive, shareable packages.

Plugin Architecture

graph TB
    A["Plugin"]
    B["Slash Commands"]
    C["Subagents"]
    D["MCP Servers"]
    E["Hooks"]
    F["Configuration"]

    A -->|bundles| B
    A -->|bundles| C
    A -->|bundles| D
    A -->|bundles| E
    A -->|bundles| F

Plugin Loading Process

sequenceDiagram
    participant User
    participant Claude as Claude Code
    participant Plugin as Plugin Marketplace
    participant Install as Installation
    participant SlashCmds as Slash Commands
    participant Subagents
    participant MCPServers as MCP Servers
    participant Hooks
    participant Tools as Configured Tools

    User->>Claude: /plugin install pr-review
    Claude->>Plugin: Download plugin manifest
    Plugin-->>Claude: Return plugin definition
    Claude->>Install: Extract components
    Install->>SlashCmds: Configure
    Install->>Subagents: Configure
    Install->>MCPServers: Configure
    Install->>Hooks: Configure
    SlashCmds-->>Tools: Ready to use
    Subagents-->>Tools: Ready to use
    MCPServers-->>Tools: Ready to use
    Hooks-->>Tools: Ready to use
    Tools-->>Claude: Plugin installed ✅

Plugin Types & Distribution

Type Scope Shared Authority Examples
Official Global All users Anthropic PR Review, Security Guidance
Community Public All users Community DevOps, Data Science
Organization Internal Team members Company Internal standards, tools
Personal Individual Single user Developer Custom workflows

Plugin Definition Structure

Plugin manifest uses JSON format in .claude-plugin/plugin.json:

{
  "name": "my-first-plugin",
  "description": "A greeting plugin",
  "version": "1.0.0",
  "author": {
    "name": "Your Name"
  },
  "homepage": "https://example.com",
  "repository": "https://github.com/user/repo",
  "license": "MIT"
}

Plugin Structure Example

my-plugin/
├── .claude-plugin/
│   └── plugin.json       # Manifest (name, description, version, author)
├── commands/             # Skills as Markdown files
│   ├── task-1.md
│   ├── task-2.md
│   └── workflows/
├── agents/               # Custom agent definitions
│   ├── specialist-1.md
│   ├── specialist-2.md
│   └── configs/
├── skills/               # Agent Skills with SKILL.md files
│   ├── skill-1.md
│   └── skill-2.md
├── hooks/                # Event handlers in hooks.json
│   └── hooks.json
├── .mcp.json             # MCP server configurations
├── .lsp.json             # LSP server configurations
├── settings.json         # Default settings
├── templates/
│   └── issue-template.md
├── scripts/
│   ├── helper-1.sh
│   └── helper-2.py
├── docs/
│   ├── README.md
│   └── USAGE.md
└── tests/
    └── plugin.test.js

LSP server configuration

Plugins can include Language Server Protocol (LSP) support for real-time code intelligence. LSP servers provide diagnostics, code navigation, and symbol information as you work.

Configuration locations:

  • .lsp.json file in the plugin root directory
  • Inline lsp key in plugin.json

Field reference

Field Required Description
command Yes LSP server binary (must be in PATH)
extensionToLanguage Yes Maps file extensions to language IDs
args No Command-line arguments for the server
transport No Communication method: stdio (default) or socket
env No Environment variables for the server process
initializationOptions No Options sent during LSP initialization
settings No Workspace configuration passed to the server
workspaceFolder No Override the workspace folder path
startupTimeout No Maximum time (ms) to wait for server startup
shutdownTimeout No Maximum time (ms) for graceful shutdown
restartOnCrash No Automatically restart if the server crashes
maxRestarts No Maximum restart attempts before giving up

Example configurations

Go (gopls):

{
  "go": {
    "command": "gopls",
    "args": ["serve"],
    "extensionToLanguage": {
      ".go": "go"
    }
  }
}

Python (pyright):

{
  "python": {
    "command": "pyright-langserver",
    "args": ["--stdio"],
    "extensionToLanguage": {
      ".py": "python",
      ".pyi": "python"
    }
  }
}

TypeScript:

{
  "typescript": {
    "command": "typescript-language-server",
    "args": ["--stdio"],
    "extensionToLanguage": {
      ".ts": "typescript",
      ".tsx": "typescriptreact",
      ".js": "javascript",
      ".jsx": "javascriptreact"
    }
  }
}

Available LSP plugins

The official marketplace includes pre-configured LSP plugins:

Plugin Language Server Binary Install Command
pyright-lsp Python pyright-langserver pip install pyright
typescript-lsp TypeScript/JavaScript typescript-language-server npm install -g typescript-language-server typescript
rust-lsp Rust rust-analyzer Install via rustup component add rust-analyzer

LSP capabilities

Once configured, LSP servers provide:

  • Instant diagnostics — errors and warnings appear immediately after edits
  • Code navigation — go to definition, find references, implementations
  • Hover information — type signatures and documentation on hover
  • Symbol listing — browse symbols in the current file or workspace

Plugin Options (v2.1.83+)

Plugins can declare user-configurable options in the manifest via userConfig. Values marked sensitive: true are stored in the system keychain rather than plain-text settings files:

{
  "name": "my-plugin",
  "version": "1.0.0",
  "userConfig": {
    "apiKey": {
      "description": "API key for the service",
      "sensitive": true
    },
    "region": {
      "description": "Deployment region",
      "default": "us-east-1"
    }
  }
}

Persistent Plugin Data (${CLAUDE_PLUGIN_DATA}) (v2.1.78+)

Plugins have access to a persistent state directory via the ${CLAUDE_PLUGIN_DATA} environment variable. This directory is unique per plugin and survives across sessions, making it suitable for caches, databases, and other persistent state:

{
  "hooks": {
    "PostToolUse": [
      {
        "command": "node ${CLAUDE_PLUGIN_DATA}/track-usage.js"
      }
    ]
  }
}

The directory is created automatically when the plugin is installed. Files stored here persist until the plugin is uninstalled.

Inline Plugin via Settings (source: 'settings') (v2.1.80+)

Plugins can be defined inline in settings files as marketplace entries using the source: 'settings' field. This allows embedding a plugin definition directly without requiring a separate repository or marketplace:

{
  "pluginMarketplaces": [
    {
      "name": "inline-tools",
      "source": "settings",
      "plugins": [
        {
          "name": "quick-lint",
          "source": "./local-plugins/quick-lint"
        }
      ]
    }
  ]
}

Plugin Settings

Plugins can ship a settings.json file to provide default configuration. This currently supports the agent key, which sets the main thread agent for the plugin:

{
  "agent": "agents/specialist-1.md"
}

When a plugin includes settings.json, its defaults are applied on installation. Users can override these settings in their own project or user configuration.

Standalone vs Plugin Approach

Approach Command Names Configuration Best For
Standalone /hello Manual setup in CLAUDE.md Personal, project-specific
Plugins /plugin-name:hello Automated via plugin.json Sharing, distribution, team use

Use standalone slash commands for quick personal workflows. Use plugins when you want to bundle multiple features, share with a team, or publish for distribution.

Practical Examples

Example 1: PR Review Plugin

File: .claude-plugin/plugin.json

{
  "name": "pr-review",
  "version": "1.0.0",
  "description": "Complete PR review workflow with security, testing, and docs",
  "author": {
    "name": "Anthropic"
  },
  "repository": "https://github.com/your-org/pr-review",
  "license": "MIT"
}

File: commands/review-pr.md

---
name: Review PR
description: Start comprehensive PR review with security and testing checks
---

# PR Review

This command initiates a complete pull request review including:

1. Security analysis
2. Test coverage verification
3. Documentation updates
4. Code quality checks
5. Performance impact assessment

File: agents/security-reviewer.md

---
name: security-reviewer
description: Security-focused code review
tools: read, grep, diff
---

# Security Reviewer

Specializes in finding security vulnerabilities:
- Authentication/authorization issues
- Data exposure
- Injection attacks
- Secure configuration

Installation:

/plugin install pr-review

# Result:
# ✅ 3 slash commands installed
# ✅ 3 subagents configured
# ✅ 2 MCP servers connected
# ✅ 4 hooks registered
# ✅ Ready to use!

Example 2: DevOps Plugin

Components:

devops-automation/
├── commands/
│   ├── deploy.md
│   ├── rollback.md
│   ├── status.md
│   └── incident.md
├── agents/
│   ├── deployment-specialist.md
│   ├── incident-commander.md
│   └── alert-analyzer.md
├── mcp/
│   ├── github-config.json
│   ├── kubernetes-config.json
│   └── prometheus-config.json
├── hooks/
│   ├── pre-deploy.js
│   ├── post-deploy.js
│   └── on-error.js
└── scripts/
    ├── deploy.sh
    ├── rollback.sh
    └── health-check.sh

Example 3: Documentation Plugin

Bundled Components:

documentation/
├── commands/
│   ├── generate-api-docs.md
│   ├── generate-readme.md
│   ├── sync-docs.md
│   └── validate-docs.md
├── agents/
│   ├── api-documenter.md
│   ├── code-commentator.md
│   └── example-generator.md
├── mcp/
│   ├── github-docs-config.json
│   └── slack-announce-config.json
└── templates/
    ├── api-endpoint.md
    ├── function-docs.md
    └── adr-template.md

Plugin Marketplace

The official Anthropic-managed plugin directory is anthropics/claude-plugins-official. Enterprise admins can also create private plugin marketplaces for internal distribution.

graph TB
    A["Plugin Marketplace"]
    B["Official<br/>anthropics/claude-plugins-official"]
    C["Community<br/>Marketplace"]
    D["Enterprise<br/>Private Registry"]

    A --> B
    A --> C
    A --> D

    B -->|Categories| B1["Development"]
    B -->|Categories| B2["DevOps"]
    B -->|Categories| B3["Documentation"]

    C -->|Search| C1["DevOps Automation"]
    C -->|Search| C2["Mobile Dev"]
    C -->|Search| C3["Data Science"]

    D -->|Internal| D1["Company Standards"]
    D -->|Internal| D2["Legacy Systems"]
    D -->|Internal| D3["Compliance"]

    style A fill:#e1f5fe,stroke:#333,color:#333
    style B fill:#e8f5e9,stroke:#333,color:#333
    style C fill:#f3e5f5,stroke:#333,color:#333
    style D fill:#fff3e0,stroke:#333,color:#333

Marketplace Configuration

Enterprise and advanced users can control marketplace behavior through settings:

Setting Description
extraKnownMarketplaces Add additional marketplace sources beyond the defaults
strictKnownMarketplaces Control which marketplaces users are allowed to add
deniedPlugins Admin-managed blocklist to prevent specific plugins from being installed

Additional Marketplace Features

  • Default git timeout: Increased from 30s to 120s for large plugin repositories
  • Custom npm registries: Plugins can specify custom npm registry URLs for dependency resolution
  • Version pinning: Lock plugins to specific versions for reproducible environments

Marketplace definition schema

Plugin marketplaces are defined in .claude-plugin/marketplace.json:

{
  "name": "my-team-plugins",
  "owner": "my-org",
  "plugins": [
    {
      "name": "code-standards",
      "source": "./plugins/code-standards",
      "description": "Enforce team coding standards",
      "version": "1.2.0",
      "author": "platform-team"
    },
    {
      "name": "deploy-helper",
      "source": {
        "source": "github",
        "repo": "my-org/deploy-helper",
        "ref": "v2.0.0"
      },
      "description": "Deployment automation workflows"
    }
  ]
}
Field Required Description
name Yes Marketplace name in kebab-case
owner Yes Organization or user who maintains the marketplace
plugins Yes Array of plugin entries
plugins[].name Yes Plugin name (kebab-case)
plugins[].source Yes Plugin source (path string or source object)
plugins[].description No Brief plugin description
plugins[].version No Semantic version string
plugins[].author No Plugin author name

Plugin source types

Plugins can be sourced from multiple locations:

Source Syntax Example
Relative path String path "./plugins/my-plugin"
GitHub { "source": "github", "repo": "owner/repo" } { "source": "github", "repo": "acme/lint-plugin", "ref": "v1.0" }
Git URL { "source": "url", "url": "..." } { "source": "url", "url": "https://git.internal/plugin.git" }
Git subdirectory { "source": "git-subdir", "url": "...", "path": "..." } { "source": "git-subdir", "url": "https://github.com/org/monorepo.git", "path": "packages/plugin" }
npm { "source": "npm", "package": "..." } { "source": "npm", "package": "@acme/claude-plugin", "version": "^2.0" }
pip { "source": "pip", "package": "..." } { "source": "pip", "package": "claude-data-plugin", "version": ">=1.0" }

GitHub and git sources support optional ref (branch/tag) and sha (commit hash) fields for version pinning.

Distribution methods

GitHub (recommended):

# Users add your marketplace
/plugin marketplace add owner/repo-name

Other git services (full URL required):

/plugin marketplace add https://gitlab.com/org/marketplace-repo.git

Private repositories: Supported via git credential helpers or environment tokens. Users must have read access to the repository.

Official marketplace submission: Submit plugins to the Anthropic-curated marketplace for broader distribution.

Strict mode

Control how marketplace definitions interact with local plugin.json files:

Setting Behavior
strict: true (default) Local plugin.json is authoritative; marketplace entry supplements it
strict: false Marketplace entry is the entire plugin definition

Organization restrictions with strictKnownMarketplaces:

Value Effect
Not set No restrictions — users can add any marketplace
Empty array [] Lockdown — no marketplaces allowed
Array of patterns Allowlist — only matching marketplaces can be added
{
  "strictKnownMarketplaces": [
    "my-org/*",
    "github.com/trusted-vendor/*"
  ]
}

Warning

: In strict mode with strictKnownMarketplaces, users can only install plugins from allowlisted marketplaces. This is useful for enterprise environments requiring controlled plugin distribution.

Plugin Installation & Lifecycle

graph LR
    A["Discover"] -->|Browse| B["Marketplace"]
    B -->|Select| C["Plugin Page"]
    C -->|View| D["Components"]
    D -->|Install| E["/plugin install"]
    E -->|Extract| F["Configure"]
    F -->|Activate| G["Use"]
    G -->|Check| H["Update"]
    H -->|Available| G
    G -->|Done| I["Disable"]
    I -->|Later| J["Enable"]
    J -->|Back| G

Plugin Features Comparison

Feature Slash Command Skill Subagent Plugin
Installation Manual copy Manual copy Manual config One command
Setup Time 5 minutes 10 minutes 15 minutes 2 minutes
Bundling Single file Single file Single file Multiple
Versioning Manual Manual Manual Automatic
Team Sharing Copy file Copy file Copy file Install ID
Updates Manual Manual Manual Auto-available
Dependencies None None None May include
Marketplace No No No Yes
Distribution Repository Repository Repository Marketplace

Plugin CLI Commands

All plugin operations are available as CLI commands:

claude plugin install <name>@<marketplace>   # Install from a marketplace
claude plugin uninstall <name>               # Remove a plugin
claude plugin list                           # List installed plugins
claude plugin enable <name>                  # Enable a disabled plugin
claude plugin disable <name>                 # Disable a plugin
claude plugin validate                       # Validate plugin structure

Installation Methods

From Marketplace

/plugin install plugin-name
# or from CLI:
claude plugin install plugin-name@marketplace-name

Enable / Disable (with auto-detected scope)

/plugin enable plugin-name
/plugin disable plugin-name

Local Plugin (for development)

# CLI flag for local testing (repeatable for multiple plugins)
claude --plugin-dir ./path/to/plugin
claude --plugin-dir ./plugin-a --plugin-dir ./plugin-b

From Git Repository

/plugin install github:username/repo

When to Create a Plugin

graph TD
    A["Should I create a plugin?"]
    A -->|Need multiple components| B{"Multiple commands<br/>or subagents<br/>or MCPs?"}
    B -->|Yes| C["✅ Create Plugin"]
    B -->|No| D["Use Individual Feature"]
    A -->|Team workflow| E{"Share with<br/>team?"}
    E -->|Yes| C
    E -->|No| F["Keep as Local Setup"]
    A -->|Complex setup| G{"Needs auto<br/>configuration?"}
    G -->|Yes| C
    G -->|No| D

Plugin Use Cases

Use Case Recommendation Why
Team Onboarding Use Plugin Instant setup, all configurations
Framework Setup Use Plugin Bundles framework-specific commands
Enterprise Standards Use Plugin Central distribution, version control
Quick Task Automation Use Command Overkill complexity
Single Domain Expertise Use Skill Too heavy, use skill instead
Specialized Analysis Use Subagent Create manually or use skill
Live Data Access Use MCP Standalone, don't bundle

Testing a Plugin

Before publishing, test your plugin locally using the --plugin-dir CLI flag (repeatable for multiple plugins):

claude --plugin-dir ./my-plugin
claude --plugin-dir ./my-plugin --plugin-dir ./another-plugin

This launches Claude Code with your plugin loaded, allowing you to:

  • Verify all slash commands are available
  • Test subagents and agents function correctly
  • Confirm MCP servers connect properly
  • Validate hook execution
  • Check LSP server configurations
  • Check for any configuration errors

Hot-Reload

Plugins support hot-reload during development. When you modify plugin files, Claude Code can detect changes automatically. You can also force a reload with:

/reload-plugins

This re-reads all plugin manifests, commands, agents, skills, hooks, and MCP/LSP configurations without restarting the session.

Managed Settings for Plugins

Administrators can control plugin behavior across an organization using managed settings:

Setting Description
enabledPlugins Allowlist of plugins that are enabled by default
deniedPlugins Blocklist of plugins that cannot be installed
extraKnownMarketplaces Add additional marketplace sources beyond the defaults
strictKnownMarketplaces Restrict which marketplaces users are allowed to add
allowedChannelPlugins Control which plugins are permitted per release channel

These settings can be applied at the organization level via managed configuration files and take precedence over user-level settings.

Plugin Security

Plugin subagents run in a restricted sandbox. The following frontmatter keys are not allowed in plugin subagent definitions:

  • hooks -- Subagents cannot register event handlers
  • mcpServers -- Subagents cannot configure MCP servers
  • permissionMode -- Subagents cannot override the permission model

This ensures that plugins cannot escalate privileges or modify the host environment beyond their declared scope.

Publishing a Plugin

Steps to publish:

  1. Create plugin structure with all components
  2. Write .claude-plugin/plugin.json manifest
  3. Create README.md with documentation
  4. Test locally with claude --plugin-dir ./my-plugin
  5. Submit to plugin marketplace
  6. Get reviewed and approved
  7. Published on marketplace
  8. Users can install with one command

Example submission:

# PR Review Plugin

## Description
Complete PR review workflow with security, testing, and documentation checks.

## What's Included
- 3 slash commands for different review types
- 3 specialized subagents
- GitHub and CodeQL MCP integration
- Automated security scanning hooks

## Installation
```bash
/plugin install pr-review

Features

Security analysis Test coverage checking Documentation verification Code quality assessment Performance impact analysis

Usage

/review-pr
/check-security
/check-tests

Requirements

  • Claude Code 1.0+
  • GitHub access
  • CodeQL (optional)

## Plugin vs Manual Configuration

**Manual Setup (2+ hours):**
- Install slash commands one by one
- Create subagents individually
- Configure MCPs separately
- Set up hooks manually
- Document everything
- Share with team (hope they configure correctly)

**With Plugin (2 minutes):**
```bash
/plugin install pr-review
# ✅ Everything installed and configured
# ✅ Ready to use immediately
# ✅ Team can reproduce exact setup

Best Practices

Do's

  • Use clear, descriptive plugin names
  • Include comprehensive README
  • Version your plugin properly (semver)
  • Test all components together
  • Document requirements clearly
  • Provide usage examples
  • Include error handling
  • Tag appropriately for discovery
  • Maintain backward compatibility
  • Keep plugins focused and cohesive
  • Include comprehensive tests
  • Document all dependencies

Don'ts

  • Don't bundle unrelated features
  • Don't hardcode credentials
  • Don't skip testing
  • Don't forget documentation
  • Don't create redundant plugins
  • Don't ignore versioning
  • Don't overcomplicate component dependencies
  • Don't forget to handle errors gracefully

Installation Instructions

Installing from Marketplace

  1. Browse available plugins:

    /plugin list
    
  2. View plugin details:

    /plugin info plugin-name
    
  3. Install a plugin:

    /plugin install plugin-name
    

Installing from Local Path

/plugin install ./path/to/plugin-directory

Installing from GitHub

/plugin install github:username/repo

Listing Installed Plugins

/plugin list --installed

Updating a Plugin

/plugin update plugin-name

Disabling/Enabling a Plugin

# Temporarily disable
/plugin disable plugin-name

# Re-enable
/plugin enable plugin-name

Uninstalling a Plugin

/plugin uninstall plugin-name

The following Claude Code features work together with plugins:

  • Slash Commands - Individual commands bundled in plugins
  • Memory - Persistent context for plugins
  • Skills - Domain expertise that can be wrapped into plugins
  • Subagents - Specialized agents included as plugin components
  • MCP Servers - Model Context Protocol integrations bundled in plugins
  • Hooks - Event handlers that trigger plugin workflows

Complete Example Workflow

PR Review Plugin Full Workflow

1. User: /review-pr

2. Plugin executes:
   ├── pre-review.js hook validates git repo
   ├── GitHub MCP fetches PR data
   ├── security-reviewer subagent analyzes security
   ├── test-checker subagent verifies coverage
   └── performance-analyzer subagent checks performance

3. Results synthesized and presented:
   ✅ Security: No critical issues
   ⚠️  Testing: Coverage 65% (recommend 80%+)
   ✅ Performance: No significant impact
   📝 12 recommendations provided

Troubleshooting

Plugin Won't Install

  • Check Claude Code version compatibility: /version
  • Verify plugin.json syntax with a JSON validator
  • Check internet connection (for remote plugins)
  • Review permissions: ls -la plugin/

Components Not Loading

  • Verify paths in plugin.json match actual directory structure
  • Check file permissions: chmod +x scripts/
  • Review component file syntax
  • Check logs: /plugin debug plugin-name

MCP Connection Failed

  • Verify environment variables are set correctly
  • Check MCP server installation and health
  • Test MCP connection independently with /mcp test
  • Review MCP configuration in mcp/ directory

Commands Not Available After Install

  • Ensure plugin was installed successfully: /plugin list --installed
  • Check if plugin is enabled: /plugin status plugin-name
  • Restart Claude Code: exit and reopen
  • Check for naming conflicts with existing commands

Hook Execution Issues

  • Verify hook files have correct permissions
  • Check hook syntax and event names
  • Review hook logs for error details
  • Test hooks manually if possible

Additional Resources