gx-arch-review

A Claude Code plugin providing slash commands for reviewing Galaxy codebase contributions.

Installation

Development/Local Testing

claude --plugin-dir /path/to/gx-arch-review

Multiple Plugins

claude --plugin-dir ./gx-arch-review --plugin-dir ./other-plugin

Usage

All commands are namespaced under gx-arch-review:. They accept input in these forms:

• Working directory path (analyze git diff)
• Git commit reference
• PR reference (e.g., #1234)
• List of Python file paths
• Planning document (analyzes files referenced in plan)

Primary Command: gx-review

/gx-arch-review:gx-review <input>

Orchestrates a comprehensive review by evaluating preconditions and launching appropriate sub-reviews. Automatically invokes relevant specialized commands based on change content.

Specialized Commands

Run individually for targeted reviews:

Command Details

gx-review (Orchestrator)

Evaluates changes and spawns sub-agents for each applicable review type. Use this as your default entry point.

py-review-code-structure

Reviews Python code for:

• Missing type annotations on public methods
• Inline imports without justification comments
• Import organization issues

py-challenge-patches

Challenges test quality by examining mocks/patches:

• Identifies tests that only test mocks
• Suggests abstractions to eliminate patching (DI, fakes, ports/adapters)
• Prefers state verification over interaction verification

gx-vitest-review

Reviews Vue/TypeScript unit tests for:

• Testing behavior vs implementation
• Proper Composition/Options API testing
• Pinia store isolation
• AI-generated test quality

gx-fastapi-review

Ensures FastAPI layer follows Galaxy patterns:

• No manual include_router calls (Galaxy auto-detects routers)
• Thin controller layer

gx-review-migration

Reviews Alembic migrations for:

• Proper use of Galaxy's migration utilities
• Migration isolation (own commit)

gx-review-model

Reviews new database models for Galaxy patterns.

gx-review-test-types

Verifies tests are in correct locations:

Common mistake: API test vs Integration test confusion

• API tests: default Galaxy config, can run against external servers
• Integration tests: require handle_galaxy_config_kwds or self._app access

review-business-logic-organization

Enforces Galaxy's three-layer architecture:

• Controllers: Thin, request/response only
• Services: Optional thin layer for API concerns
• Managers: Business logic lives here
• Models: Database access only

review-di

Enforces dependency injection patterns:

• Galaxy-style depends() over FastAPI Depends()
• Constructor injection over app property access
• StructuredApp interface over UniverseApplication
• @galaxy_task for Celery tasks

Plugin Structure

gx-arch-review/
├── .claude-plugin/
│   └── plugin.json      # Plugin manifest (required)
├── commands/            # Slash commands
│   ├── gx-review.md
│   ├── py-review-code-structure.md
│   ├── gx-vitest-review.md
│   └── ...
└── README.md

Relationship to galaxy-architecture Repository

This plugin is built from the galaxy-architecture repository. See the Agentic Code Review section in the root README for:

• Build infrastructure in review/
• The feedback loop between documentation, commands, and reviews
• How generated commands improve architecture documentation

Commands are:

• Static: Hand-written in review/static_commands/
• Generated: Created from architecture topics via /generate-agentic-op

Contributing

To add a new slash command:

1. Create static_commands/gx-review-<name>.md with:

   ---
   description: Brief description for command listing
   ---
   
   # Command Title
   
   Input format description...
   Review criteria...
   Code patterns (good/bad examples)...
   Output format...

2. Add precondition entry in static_commands/gx-review.md
3. Run make to publish
4. Test with real Galaxy PRs/commits

License

Same license as the Galaxy project.