Rust SDK

Application code typically depends on xlai-native (or xlai-wasm in the browser). Both expose RuntimeBuilder, chat and agent sessions, and backend configuration types.

Platform entrypoints

Crate	Use when
xlai-native	Native binaries and servers on macOS, Linux, or Windows. Optional llama and qts features.
xlai-wasm	wasm32-unknown-unknown builds and JavaScript interop.
xlai-ffi	C ABI / shared library embedding (wraps xlai-native).

Domain types and traits live in xlai-core (semver-stable on crates.io). Session APIs live in xlai-runtime and are re-exported through xlai-native / xlai-wasm. The xlai-facade crate is an internal workspace helper (not on crates.io) used only by xlai-native for native aggregate wiring; xlai-wasm does not depend on it.

Minimal pattern

1. Build a RuntimeBuilder and register a chat backend (OpenAiConfig, LlamaCppConfig, …).
2. Call .build()? to obtain a runtime.
3. Use chat_session() or agent_session()? for typed conversations.

Concrete examples (OpenAI, llama.cpp, streaming agents) are in the repository README — they are kept there so they stay in sync with the workspace.

Topics

• Chat, agents, and tools — tool registration, execution modes, and the agent tool loop
• Streaming — ChatChunk and ChatExecutionEvent behavior

Optional: QTS in native builds

Enable the qts feature on xlai-native when you need QtsTtsModel without taking the dependency in every build.