- How do I install Laravel Model Explorer without breaking my existing frontend stack (Inertia/Livewire/Blade)?
- The package includes a self-contained Vue 3 SPA embedded via Vite, so it won’t conflict with your existing frontend. Just run `composer require --dev onelearningcommunity/laravel-model-explorer`, and access it at `/_model-explorer`. No `vendor:publish` is needed unless you customize paths or middleware.
- Does this package support Laravel 10 or older versions?
- No, it requires Laravel 11, 12, or 13 and PHP 8.3+. Older versions may need polyfills, but the package is optimized for modern Laravel stacks. Check the [GitHub](https://github.com/One-Learning-Community/laravel-model-explorer) for updates.
- Can I restrict access to the Model Explorer in production?
- Yes. Disable it entirely with `MODEL_EXPLORER_ENABLED=false` in your `.env`, or use the `viewModelExplorer` gate to restrict access. By default, it’s only available locally, making it safe for production.
- How does the relationship graph perform with deeply nested or large schemas (e.g., 500+ models)?
- The graph uses a force-directed SVG layout, which can become slow with large schemas. For now, there’s no lazy-loading or pagination, but the package avoids runtime overhead by leveraging Eloquent metadata. Test with your schema size before relying on it for complex DDD architectures.
- Will sensitive data (e.g., passwords, API keys) in model accessors or `toArray()` be exposed?
- The package doesn’t filter accessor output, so sensitive data exposed via `toArray()` or accessors will appear in the UI. Use Laravel’s `hidden()` or `visible()` arrays to protect fields, or restrict access via middleware if needed.
- Can I customize the UI (e.g., add filters, change layouts) without forking the package?
- The package is designed to be extended but intentionally locked down to avoid breaking changes. You can publish the config to override middleware or model paths, but UI customizations would require forking or contributing back. The Vue 3 SPA is scoped to `#app` for isolation.
- How does the MCP server work for AI agents like Claude Code or Cursor?
- The package includes a local MCP server (`model-explorer`) exposing five tools: `list-models`, `inspect-model`, `find-model`, `model-source`, and `model-neighbors`. AI agents can query live model structure, traits, and relationships without scanning your source code or relying on stale context.
- Does this package work with custom database columns (e.g., JSON, UUIDs, or enum types)?
- It relies on `spatie/laravel-model-info` for column metadata, which supports MySQL, PostgreSQL, and SQLite. Custom types like UUIDs or JSON columns should work, but edge cases (e.g., complex JSON structures) may require manual overrides in the config.
- How do I handle the `npm run build` step in CI/CD to keep PHP and JS in sync?
- The package includes a pre-built Vue 3 SPA, but you should run `npm install && npm run build` in your CI pipeline to ensure assets are up to date. The package’s release workflow doesn’t handle this automatically, so integrate it into your build process.
- Are there alternatives to this package for exploring Eloquent models?
- For UI-based exploration, alternatives include Laravel Scout’s model introspection (limited) or custom admin panels. For AI integration, the MCP server is unique—most packages lack real-time model introspection. If you need a lighter option, consider `spatie/laravel-model-info` alone for metadata, but it lacks the UI and graph visualization.