Experimental Community Project: This repository is maintained by Shopware Labs and is not an official Shopware product. It is not affiliated with, endorsed by, or sponsored by Anthropic or any other AI provider. "Claude" and "Claude Code" are trademarks of Anthropic. This project is provided as-is without warranty.
A Claude Code plugin marketplace for Shopware development. Provides development tools, test generation, code research, and more — integrated directly into your Claude Code workflow.
Requirements: Claude Code installed.
Add the marketplace, then install the plugins you need:
/plugin marketplace add shopwareLabs/ai-coding-tools
/plugin install dev-tooling@shopware-ai-coding-toolsRestart Claude Code after installing plugins that include MCP servers.
These plugins work best alongside a few Claude Code tweaks. Turn on ENABLE_TOOL_SEARCH=1 for deferred MCP tool loading. Add complementary marketplaces like Anthropic's superpowers and it-bens/ai-tools (for prompt-engineering and redundant-read-blocker). Pre-approve common tools in settings.json.
👉 See docs/claude-code-setup.md for the full recipe.
| Plugin | Description | Components |
|---|---|---|
| dev-tooling | PHPStan, ECS, PHPUnit, ESLint, Stylelint, Jest, and more via MCP servers. Includes Shopware LSP. | 🔌 MCP · 🪝 Hooks · 🎯 Skills · 🧠 LSP |
| gh-tooling | GitHub CLI wrapper for PRs, issues, CI runs, and search. | 🔌 MCP · 🪝 Hooks · 🎯 Skills |
| test-writing | Automated PHPUnit test generation and validation for Shopware 6. | 🎯 Skills · 🤖 Agents · 🔌 MCP |
| chunkhound-integration | Semantic code research using ChunkHound. | 🔌 MCP · 🎯 Skills · 🤖 Agents · 💬 Cmds · 🪝 Hooks |
| ci-failure-interpretation | CI failure log interpretation for GitHub Actions workflows. | 🎯 Skills |
| contributor-writing | ADRs, PR descriptions, commit messages, RELEASE_INFO, and UPGRADE entries for the Shopware core repository. | 🎯 Skills |
| code-contribution-analysis | Analyzes GitHub pull requests and issues in depth. Two skills fetch contribution data and research architectural context via ChunkHound. | 🎯 Skills |
Three MCP servers for PHP and JavaScript operations plus Shopware LSP for intelligent code completion. Supports native, Docker, Vagrant, and DDEV environments.
/plugin install dev-tooling@shopware-ai-coding-tools- PHP: PHPStan static analysis, ECS code style, PHPUnit test runner with coverage gap analysis, Symfony Console
- Administration JS: ESLint, Stylelint, Prettier, Jest, TypeScript, Vite builds
- Storefront JS: ESLint, Stylelint, Jest, Webpack builds
- Shopware LSP: Service ID completion, Twig templates, snippets, routes, feature flags
After installing, ask Claude to help you set up the plugin — the setting-up skill will walk you through configuration. Prerequisites: jq, restart after install. For LSP: shopware-lsp binary in PATH.
See full documentation for configuration and tool reference.
GitHub CLI MCP server for pull requests, issues, CI runs, jobs, commits, and search. Works without configuration when gh is authenticated.
/plugin install gh-tooling@shopware-ai-coding-tools- PRs: view, diff, list, checks, comments, reviews, files, commits
- Issues: view, list
- CI: run status, logs, job-level debugging, annotations
- Other: commit-to-PR lookup, cross-repo search, raw API access
After installing, ask Claude to help you set up the plugin — the setting-up skill will check prerequisites and optionally create a config file. Prerequisites: jq, gh CLI authenticated, restart after install.
See full documentation for configuration and tool reference.
Generates and validates PHPUnit unit tests for Shopware 6. Analyzes source classes, detects the test category (DTO, Service, Flow/Event, DAL, Exception), generates tests, reviews them against 46 Shopware-specific rules, and iterates fixes until they pass. Also supports team-based consensus review using Agent Teams (experimental).
/plugin install test-writing@shopware-ai-coding-toolsJust ask Claude to generate tests — the skill activates automatically:
Generate unit tests for src/Core/Content/Product/ProductEntity.php
Prerequisites: dev-tooling plugin installed, .mcp-php-tooling.json in project root, restart after install.
See full documentation for categories, rules, and workflow details.
Semantic code research using ChunkHound's multi-hop search and LLM synthesis. Understands code architecture, traces data flows, and discovers component relationships.
/plugin install chunkhound-integration@shopware-ai-coding-tools/research how does authentication work in this codebase?
/research find all payment service dependencies
After installing, ask Claude to help you set up the plugin — the setting-up skill will guide you through ChunkHound installation, embedding provider configuration, and indexing. Restart after install.
See full documentation for setup and configuration.
Knowledge skill for interpreting CI failure logs from Shopware GitHub Actions workflows. Covers PHPUnit, PHPStan, ECS, ESLint, TypeScript, Stylelint, Prettier, Jest, Playwright, ludtwig, and Lighthouse.
/plugin install ci-failure-interpretation@shopware-ai-coding-toolsThe skill activates automatically when analyzing CI failures — just ask Claude to interpret logs or debug a failed CI run. No prerequisites beyond installation.
See full documentation for supported tools and failure patterns.
Writing skills for Shopware core contributors: Architecture Decision Records, PR descriptions, commit messages, and RELEASE_INFO/UPGRADE entries. Analyzes branch diffs, classifies changes, asks for context, and writes content calibrated to change magnitude.
/plugin install contributor-writing@shopware-ai-coding-toolsWrite an ADR about switching to Redis for cart persistence
Write a PR description for my changes
Write a release info entry for my changes
Generate a squash commit message for this branch
Write a commit message for my changes
Skills activate automatically. Requires gh-tooling plugin for PR analysis.
See full documentation for workflow details and writing rules.
Two skills for analyzing GitHub pull requests and issues in depth. Each fetches the contribution data and researches its architectural context, producing a structured analysis covering scope, impact, and code relationships.
/plugin install code-contribution-analysis@shopware-ai-coding-tools- PR analysis: metadata, diff, files, reviews, and inline comments fetched from GitHub; architectural impact researched via
chunkhound-integrationusing an incremental (Stage 1/2/3) strategy - Issue analysis: metadata and comments fetched from GitHub; affected code area located and researched via
chunkhound-integration - Standalone and embeddable: works in interactive Claude Code sessions and as a building block for Claude Agent SDK applications (e.g., automated briefing tools)
Prerequisites: chunkhound-integration plugin installed and configured. The skills stop with an error if ChunkHound is not available at runtime. GitHub access (MCP server, gh CLI, or API) is fetched with whatever is available in the session — no specific tool required.
Just ask Claude to analyze a PR or issue and the right skill activates:
Analyze PR #4521 in shopware/shopware
Analyze issue #8910 in shopware/shopware
See full documentation for details and Agent SDK integration examples.
Selected skills are available as portable packages following the Agent Skills specification. Download ZIPs from the latest release and install them in your tool of choice:
To opt a skill in for export, place an empty .agent-skills file next to its SKILL.md.
Found a bug or quality issue? Open an issue using our specialized templates for commands, skills, agents, hooks, MCP servers, or other components.
Some plugins in this marketplace integrate with external services (e.g. chunkhound-integration wraps the ChunkHound CLI with user-configured embedding providers such as VoyageAI, OpenAI, or Ollama). These integrations are entirely opt-in and user-configured: the plugins invoke user-installed CLI tools using environment variables and configuration files supplied by the user. Shopware Labs does not receive, store, or route any data processed through these integrations. Users who configure external AI providers are solely responsible for compliance with those providers' terms of service and any applicable data protection requirements, including obligations arising from transfers to third countries.
This project is licensed under the MIT License.
Note
Yes, an AI wrote this README. And everything else as well. Yes, a human told it to add emojis. The human has ADHD, which — as it turns out — means his brain already ran on associative pattern-matching and nonlinear leaps before LLMs made it cool. They call him ... LLMartin. The emojis are a feature.