How enzoscarpa-fnk/customllm works
Project Type and Purpose
• This is a full-stack AI chat web application (“CustomLLM”) built on Laravel 12 with an Inertia/Vue 3 front end and Tailwind CSS.
• It provides authentication (single-user and teams via Fortify & Jetstream), persistent conversation history, real-time streaming from various large language models (OpenAI’s GPT, Google Gemini, Anthropic Claude, etc.), and personalized instruction panels.
File Structure Analysis
• Top level:
• config/: Laravel configuration (DB, queues, OpenAI)
• app/: core Laravel MVC code (Controllers, Models, Services, Fortify/Jetstream integration)
• resources/js/: Vue 3 components, pages, composables, Inertia setup
• routes/: web.php & api.php define application endpoints
• database/: migrations, factories, seeders
• public/: index.php entry point + static assets
• storage/: cache, sessions, file uploads
• tests/: feature & unit tests
• Architectural patterns:
• MVC in Laravel (Controllers → Services → Models → DB)
• Inertia bridging Laravel to a SPA-style Vue front end
• Service layer (app/Services/ChatService) orchestrating API calls
• Queued jobs and streaming to support real-time token delivery
README Insights
• Confirms core tech stack: Laravel, Inertia, Vue Composition API, Tailwind
• Highlights real-time streaming (via a Composable + ChatService)
• Mentions model selector (stateful per-chat model choice) and secure, collaborative features (Fortify & Jetstream)