@openvole/paw-ollama
v1.0.1
Published
Brain Paw powered by Ollama for local LLM inference
Downloads
488
Readme
@openvole/paw-ollama
Brain Paw powered by Ollama for local LLM inference.
Part of OpenVole — the microkernel AI agent framework.
Install
npm install @openvole/paw-ollamaConfig
{
"name": "@openvole/paw-ollama",
"allow": {
"network": ["127.0.0.1"],
"env": ["OLLAMA_MODEL", "OLLAMA_HOST"]
}
}Environment Variables
| Variable | Description |
|----------|-------------|
| OLLAMA_MODEL | Model name to use (e.g. llama3) |
| OLLAMA_HOST | Ollama server address (default http://127.0.0.1:11434) |
Brain
Implements the Think phase — receives the conversation context and registered tools, returns the next assistant message and any tool calls. Uses a locally running Ollama instance.
BRAIN.md
The system prompt is loaded from .openvole/paws/paw-ollama/BRAIN.md. On first startup, the paw scaffolds a default prompt there. Edit this file to customize how the Brain reasons and responds.
The old .openvole/BRAIN.md path is no longer used.
