Jarvis — a self-hosted AI assistant

Jarvis is a self-hosted personal AI assistant split between a home Mac Mini — the brain: gateway, memory, local models, voice, code sandbox, skills — and a VPS for public ingress, connected over a private Tailscale network. Local-first is the point: a Qwen3 model served through Ollama, a pgvector memory store with bi-temporal corrections (facts are superseded, never deleted), and a wake-word-to-speech voice loop with barge-in all run on hardware at home.

The keystone is the contracts package. Every cross-service boundary is a typed, versioned Pydantic model with generated TypeScript, a runnable deterministic mock, and a contract test suite that any real implementation must pass — so seven service tracks build in parallel against mocks and swap in one at a time. The trickiest machinery is the approval gate: a LangGraph interrupt pauses a streaming turn for a human decision, resolvable over WebSocket or REST with first resolution winning, and durable enough to survive a process restart mid-approval.