Local LLMs for Laravel through OpenAI-compatible APIs (Ollama, vLLM, LM Studio, llama.cpp): providers and purpose routes in the database with fallback, versioned prompt templates, structured output validated against a JSON schema, a call log with pruning, token quotas, a process-engine effect and a test fake.

Process-engine screens for dskripchenko/laravel-admin: instances, live flow diagram, journal, incidents and operator actions.

Laravel bridge for dskripchenko/process-engine: database store, queue workers, scheduler, artisan commands, a Processes facade, Eloquent model binding and state projection.