Solo.io has open-sourced two projects to address enterprise challenges with agentic AI in production: Agent Registry and Agent Evals. Agent Registry, donated to the CNCF as a sandbox project, provides a central hub for governing and discovering AI agents, MCP tools, and agent skills across frameworks like Kubernetes. Agent Evals benchmarks agent reliability using OpenTelemetry-based metrics and tracing to score performance and determine the level of human oversight needed. A third tool, Agent Gateway, handles security and policy enforcement for agent-to-LLM and agent-to-MCP traffic, including sensitive data redaction and integration with Istio Ambient mode. Together, these tools aim to make agentic AI deployments governable, reliable, and secure for enterprise production use.
Sort: