Show HN: Recursant, a mesh-based control plane for AI agents (github.com)

Recursant, an open-source platform designed for managing AI agents, was announced as a solution to the control challenges faced by enterprises scaling AI deployments. It functions as an "Istio for AI agents," providing a governance and observability framework that addresses crucial issues such as inter-agent communication, compliance, and security. This platform enables agents to operate effectively across various stacks and clouds by creating a robust agentic mesh architecture, featuring both a control plane and a data plane. The technical framework of Recursant comprises a control plane built with Flask and React, supported by PostgreSQL and Kafka, which acts as the centralized management hub for agent metadata and policies. The data plane utilizes a Python sidecar process, managing inter-agent traffic using mutual TLS (mTLS) and the A2A protocol. This structure not only enhances security and compliance but also allows for detailed monitoring and auditing—ensuring that agent interactions are both safe and accountable. With Recursant, enterprises can deploy AI agents at scale without needing to rewrite existing code, addressing critical questions around access control, data leaks, and decision-making transparency.