local_ai_kit 0.0.3

LocalAI Kit facade: LocalAI.initialize wiring, facades (llm/stt/tts/voice/ models/runtime), model download manager with resumable downloads, LRU runtime scheduler, voice pipeline with barge-in, typed [...]

Changelog

0.0.3

• Add llama.cpp-backed LLM and embedding support to the facade and model hub.
• Support registering external, bring-your-own model files.
• Expand the bundled catalog with GGUF, Moonshine v2 and Dolphin models.

0.0.2

• Extract sentence boundaries from streaming LLM text and pipeline responses into per-sentence TTS for lower time-to-first-audio.
• Cancel barge-in before stopping output to avoid audio glitches.
• Introduce a Model Context Protocol (MCP) plugin architecture.
• Coordinated version bump; correct the bundled LICENSE text.

0.0.1

• Initial public release.