Core
| Feature | Status |
|---|---|
| LLM inference (llama.cpp via FFI) | ✅ Done |
| Default tools: shell, read, write, glob, websearch | ✅ Done |
| MCP server support | ✅ Done |
| REST API | 🚧 Planned |
Orchestrator
A multi-agent orchestration layer allowing multiple specialized workers to collaborate on complex tasks.| Feature | Status |
|---|---|
| Model routing between workers | 🚧 Planned |
| Per-worker system prompts | 🚧 Planned |
| Per-worker built-in tool sets | 🚧 Planned |
Voice Mode
Fully local speech interaction using embedded inference for all stages.| Feature | Status |
|---|---|
| Voice Activity Detection (VAD) inference | 🚧 Planned |
| Noise Cancellation inference | 🚧 Planned |
| Speech-to-Text (STT) inference | 🚧 Planned |
| Text-to-Speech (TTS) inference | 🚧 Planned |
Multimodal
| Feature | Status |
|---|---|
| Image generation model inference | ✅ Done |
| Video generation model inference | 🚧 Planned |
Contributing
Want to help build these features? Contributions are welcome.- Open a pull request to contribute code
- Open an issue to report bugs or propose ideas