gx-arch-review

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	Focus Area	Precondition
/gx-arch-review:py-review-code-structure	Type annotations, import organization	Python code
/gx-arch-review:py-challenge-patches	Mock/patch quality in tests	Python tests
/gx-arch-review:gx-vitest-review	Frontend unit test quality	Client test code (*.test.js/ts)
/gx-arch-review:gx-fastapi-review	FastAPI layer patterns	API endpoint changes
/gx-arch-review:gx-review-migration	Alembic migration patterns	Database migrations
/gx-arch-review:gx-review-model	Database model patterns	New database models
/gx-arch-review:gx-review-test-types	Test placement verification	Test files
/gx-arch-review:review-business-logic-organization	Controller/Service/Manager layers	API endpoint changes
/gx-arch-review:review-di	Dependency injection patterns	Python code

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:

Test Type	Location	When to Use
Unit	test/unit/	No running server needed
API	lib/galaxy_test/api/	Server needed, default config
Integration	test/integration/	Server + custom config
Selenium	lib/galaxy_test/selenium/	Browser automation

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