claude-plugins CLAUDE.md

CLAUDE.md

This file provides guidance to Claude Code (claude.ai/code) when working with code in this repository.

What This Is

A Claude Code plugin marketplace repository (stilero/claude-plugins). Contains multiple independent plugins distributed via /plugin marketplace add stilero/claude-plugins. There is no build system, package manager, or test suite — plugins are pure markdown/JSON.

Repository Structure

plugins/
  <plugin-name>/
    .claude-plugin/
      plugin.json        # Required manifest (name, description, version, author)
    commands/            # Slash commands (markdown files)
    agents/              # Subagent definitions (markdown files)
    skills/              # Skill definitions (SKILL.md with YAML frontmatter)
      <skill-name>/
        SKILL.md
        references/      # Optional supporting docs
    .mcp.json            # Optional MCP server config
    README.md

Current Plugins

• pr-comment-fixer — Fetches and fixes unresolved GitHub PR review comments. Has a skill and a slash command.
• hardcore-code-reviewer — Strict code reviewer that spawns 12 parallel subagents (bug-hunter, security-reviewer, architecture-reviewer, test-reviewer, silent-failure-hunter, performance-reviewer, complexity-reviewer, observability-reviewer, api-contract-reviewer, data-migration-reviewer, accessibility-reviewer, type-safety-reviewer). Has a skill, a command, and 12 agent definitions.
• shopify-apps — Guidance for Shopify app development. Has a skill with reference docs.
• github-issue-generator — Evaluates conversation findings and creates/updates GitHub issues with duplicate checking.
• obsidian-knowledge-bank — Persistent indexed knowledge bank for learnings, ADRs, technical context, and session notes.
• tech-debt-audit — Spawns 11 parallel agents to audit a codebase for technical debt, groups findings into actionable issues, and offers GitHub issue creation.

Plugin Conventions

• plugin.json is the only required file — it lives in .claude-plugin/.
• Skills use YAML frontmatter (name, description) in SKILL.md. The description field is critical for triggering — it must clearly describe when the skill should activate.
• Agent markdown files define subagent behavior with YAML frontmatter specifying tools, model, and description.
• Commands are markdown files with YAML frontmatter (description, optional allowed-tools).
• No code compilation or dependencies. All plugin logic is expressed as markdown prompts.