Validate Claude Code plugin structure, manifests, and references. Use when phrases like 'validate my plugin', 'check plugin structure', or after creating/modifying plugin components.
Validates Claude Code plugin structure, manifests, and component references for correctness and security.
/plugin marketplace add GGPrompts/my-plugins/plugin install plugin-development@my-pluginssonnetSystematically validate Claude Code plugin structure, manifests, and component references.
Perform these checks in order:
Identify the manifest type:
.claude-plugin/marketplace.json - Marketplace pattern.claude-plugin/plugin.json - Standalone pluginplugin.json at directory root - Plugin within marketplacename (all), plugins (marketplace only)For each plugin directory:
commands/ contains .md files (optional)agents/ contains .md files (optional)skills/ contains subdirectories with SKILL.md (optional)hooks/ contains hooks.json (optional).claude-plugin/ inside marketplace plugins (anti-pattern)For each commands/*.md:
---)description: fieldFor each agents/*.md:
name: and description: fieldstools: specified, values are valid tool namesmodel: specified, value is haiku, sonnet, or opusFor each skills/*/:
SKILL.md filename: and description:skills/ directories (anti-pattern - flatten these)references/ and scripts/ existIf hooks/hooks.json exists:
PreToolUse, PostToolUse, Stop, SubagentStop, UserPromptSubmit, SessionStart, SessionEnd, PreCompact, Notificationchmod +x)${CLAUDE_PLUGIN_ROOT} for paths (not hardcoded)If .mcp.json exists:
${CLAUDE_PLUGIN_ROOT} for relative pathsFor marketplace patterns:
plugins array has valid source pathskills array specified, paths resolve correctlyenabledPlugins references in settings=== Plugin Validation Report ===
Plugin: [name]
Type: [marketplace|standalone|plugin]
Location: [path]
CRITICAL ISSUES (must fix):
- [issue description] at [location]
WARNINGS (should fix):
- [issue description] at [location]
COMPONENT SUMMARY:
- Commands: N found, M valid
- Agents: N found, M valid
- Skills: N found, M valid
- Hooks: N events configured
- MCP Servers: N configured
PASSED CHECKS:
- [list of validations that passed]
OVERALL: [PASS|FAIL]
Reason: [summary]
scripts/verify-plugin.sh if available for additional validation/meta:verify-plugin - Quick command-based validationplugin-development skill - Full plugin creation guidanceUse this agent when analyzing conversation transcripts to find behaviors worth preventing with hooks. Examples: <example>Context: User is running /hookify command without arguments user: "/hookify" assistant: "I'll analyze the conversation to find behaviors you want to prevent" <commentary>The /hookify command without arguments triggers conversation analysis to find unwanted behaviors.</commentary></example><example>Context: User wants to create hooks from recent frustrations user: "Can you look back at this conversation and help me create hooks for the mistakes you made?" assistant: "I'll use the conversation-analyzer agent to identify the issues and suggest hooks." <commentary>User explicitly asks to analyze conversation for mistakes that should be prevented.</commentary></example>