Skip to main content

​
Overview

Nevron is Neurobro’s open-source agent framework: the foundation the Neurobro AI agents are built on. It provides the building blocks for autonomous agents - planning, memory, tool use, and multi-model LLM integration - packaged as a lightweight, modular Python framework.

Documentation

Technical documentation

GitHub

Source code

​
Key traits

  • Lightweight - minimal dependencies and a small footprint.
  • Modular - planning, memory, feedback, and tool modules can be swapped or extended.
  • Autonomous - agents make their own decisions via a planning and feedback loop.
  • Multi-model - pluggable LLM providers, chosen per use case.
  • Extensible memory - vector stores for agent context and recall.
  • Easy to deploy - runs anywhere Python and Docker do.

​
What’s included?

  • Broad tool integrations (Discord, Slack, WhatsApp, Google Drive, and more).
  • Multiple LLM providers, including open-source models.
  • Vector-store memory backends (Pinecone, Weaviate).
  • A planning module with dynamic tool selection and feedback loops.

​
Status

Public development of Nevron is currently paused. The framework has a stable release and continues to serve as a core component of Neurobro; ongoing work happens privately rather than in the public repository.