🎯High FreedomSwap engine, pipeline, parser, or formatter. Local llama.cpp, cloud APIs, or your own backend – mix freely, no lock-in.
🧩Highly ExtensibleEverything is a trait. Implement what you need, skip the rest. A thin core that stays out of your way.
⚡High PerformanceRust + Tokio async. Arc-shared blueprints, zero-cost cloning. Parallel tool execution. No GC, no surprises.
🌐Cross-PlatformOne codebase. Native binaries on Windows/Linux/macOS, or WASM in the browser. Runs wherever Rust runs.