smooai-smooth-operator-server 1.4.0

Reference WebSocket service for smooth-operator — speaks the schema-driven protocol over a smooth-operator KnowledgeChatRuntime.
Documentation

smooai-smooth-operator-server

The reference WebSocket service for smooth-operator — the deployment surface.

It speaks the schema-driven protocol in spec/ over a smooth-operator-backed KnowledgeChatRuntime, so the generated TypeScript / Python / Go / .NET / Rust clients connect and drive real LLM turns unmodified. Run it locally with cargo run -p smooai-smooth-operator-server (in-memory storage, no database to provision), or deploy it three ways from the one codebase:

  • Local — cargo run, in-memory adapter, for development and the cross-language E2E.
  • Kubernetes — Helm + ArgoCD, Postgres + pgvector, the Redis/NATS backplane for multi-pod scale-out.
  • AWS serverless (SST) — the companion smooth-operator-lambda crate behind API Gateway WebSocket, with DynamoDB + S3 Vectors.

See docs/DEPLOY.md for the full matrix.

Part of the smooth-operator service — Smoo AI's polyglot AI agent service (one schema-driven WebSocket protocol, five native clients, AWS-serverless or Kubernetes deploy). See the repository for the architecture, protocol, and the other crates.

License

MIT © 2026 Smoo AI. See LICENSE.