Docker service stack for a comprehensive local ai experience, running on a normal home hardware setup with single GPU. Project-Status: Working, Backlog.

A LiteLLM proxy server with built-in message sanitization for Mistral models

OpenAI-compatible proxy that makes local LLM tool calling actually work — schema validation, retry with feedback, model escalation, and context condensing

EU-sovereign multi-model LLM serving pipeline — Apertus 70B + Devstral 24B + EuroLLM 22B with EU AI Act Art. 52/53 transparency

Autonomous dev agent that runs on your hardware. No cloud. No API keys. Ships code while you sleep.

27 one-shot FR agents on Mistral La Plateforme — Large, Ministral, Devstral, Pixtral, zero-429