...Behind the scenes, it leverages Mistral’s coding-optimized LLM stack (including models tuned for code understanding and generation), with project-wide context awareness: it scans your file structure, Git status, and recent history to inform suggestions so that generated code aligns with existing context.