Public development of Nevron is paused. The open-source release is stable and stays available, while Neurobro’s work on its own agents happens privately.
Documentation
Technical documentation
GitHub
Source code
What’s included
- Planning - agents choose their next step through a planning and feedback loop, with dynamic tool selection.
- Memory - vector-store backends such as Pinecone and Weaviate for context and recall.
- Multiple LLM providers - pluggable models, including open-source ones, chosen per use case.
- Tool integrations - Discord, Slack, WhatsApp, Google Drive, and more.
- Modular - swap or extend the planning, memory, feedback, and tool modules.
- Lightweight and portable - minimal dependencies, and it runs anywhere Python and Docker do.