RUST AI GATEWAY / OPEN SOURCE
Xiaomaolv小毛驴
A production-minded AI gateway built in Rust. Plug in your provider, select a channel, and ship fast with streaming replies, extensible plugins, and reliable local-first runtime patterns.
OpenAI-CompatibleTelegram + HTTPStreaming RepliesPlugin Architecture
Provider Abstraction
MiniMax / OpenAI / Compatible
Channel Modes
Telegram Polling + Webhook
Memory
SQLite + Hybrid Sidecar
Why This Gateway Feels Different
Rust Performance
Built for low-latency message routing and predictable runtime behavior.
Streaming UX
Telegram replies stream incrementally so users see answers earlier.
Safety-First Defaults
Polling mode works locally and avoids webhook friction during setup.
Flexible Memory
Start with SQLite-only, then switch to hybrid memory when needed.
Plugin Extensions
Custom provider and channel registries let you evolve architecture safely.
MVP in One Command
Run MiniMax + Telegram with a script and ship a real integration quickly.
Provider Layer
Plug any OpenAI-compatible endpoint into a unified provider interface without touching channel logic.
Channel Layer
Route messages through HTTP or Telegram with polling/webhook runtime modes and streaming output support.
Memory Layer
Keep it simple with SQLite-only memory, then scale retrieval strategy with a zvec sidecar when your use case grows.
Quick Start Flow
The recommended setup is MiniMax + Telegram. Configure two environment variables, run one script, and start chatting with your bot.
$ cp .env.realtest.example .env.realtest
$ # Fill MINIMAX_API_KEY + TELEGRAM_BOT_TOKEN
$ ./scripts/run_mvp_minimax_telegram.sh
$ curl -sS http://127.0.0.1:8080/health
FAQ
Build Bot Infrastructure at Gateway Speed
If you want a practical and extensible Rust foundation for AI channels, xiaomaolv is ready to be your control plane.