Praval can be used to build a multi-agent system today with native agent-to-agent communication. Agents can be imbued with memory (native Chroma DB integration) and the Reef communication sub-system is akin to a message queue. Reef has native support for RabbitMQ as well. Observability features are being built with Open Telemetry compatibility. It supports OpenAI LLMs as of now, and I'm hoping to build support for Anthropic, Cohere, and other models via OpenRouter and other providers soon.
I'm building a couple of apps on top of Praval - one of them is Praval Deep Research: https://github.com/aiexplorations/praval_deep_research - this is a local-first, agentic research tool for people using Arxiv to do research and ask questions to understand papers. Can help you search, index and do Q&A on top of research papers, and also get prompted on research topics.
Praval is on PyPI: https://pypi.org/project/praval/
There's a site with documentation here: https://pravalagents.com/
Still an early project and I'm sure there's lots of scope for new features. Give it a spin and let me know how you like it.