Oryxis is a high-performance, Rust-based personal AI assistant (Jarvis-inspired) designed to bridge the gap between Large Language Models and system-level execution.
Unlike standard chatbots, Oryxis is built for speed, memory, and action. It doesn't just talk; it thinks and interacts with your system.
[!IMPORTANT] Note for Hack Club Testers: > When you launch the application for the first time, please type "tester" in response to the first question. This will automatically provision a temporary API key for your testing session.
Oryxis is built on three main pillars:
- Thinking Cycle: A decision-making loop that analyzes user intent, queries long-term memory, and selects the most appropriate "skill" (LLM-driven tool use).
- Long-Term Memory Integration: ** Coming soon... **
- Search & Retrieval: A smart skill-fetching mechanism that keeps the LLM's context window clean by only loading relevant tools on demand.
- Gemini
- Groq
- LLMAPI.ai
- Rust (latest stable)
- EdgeTTS python package (latest)
Create a .env file in the 'Oryxis' directory as the executable:
API_TYPE=LLMAPI // GROQ or GEMINI
API_KEY=Your Api key
LLM_MODEL=gemini-2.5-flash // that line depends on api type.
TTS=en-AU-WilliamMultilingualNeural // Edge-tts- You can write it python module like style or which language dou yo want (it have to can crate a .dll file and stable with ctypes)
- If your skill is not a .py and its .dll, you have to write python bridge like 'ORYXIS/skills/lib/*.py files'
- Create a good .yaml file (Oryxis/skills/*.yaml) and add your skill into skill index file (ORYXIS/memory/skills_index.json)
- When you are done with configs and skills you can run oryxis now! (you have to build .dll files also look 'libraries_opensource/' for it and compile skills, then move .dll files into 'ORYXIS/skills/lib')
- cd ORYXIS
- cargo run --release
- prompt is made by me and for me, so you can see a name or personalized texts in prompt file you have to change it for your self
- also, read whole prompt because of there is some rules can be bad for your self.