WSO2 Agent Manager

WSO2 Agent Manager is an open-source, Apache 2.0 control plane that deploys, observes, and governs AI agents across frameworks, models, and environments, with verifiable agent identity, 40-plus guardrails mapped to the OWASP Top 10, and a Kubernetes-native sandboxed runtime.

WSO2 Agent Manager is an open enterprise control plane built to rein in AI agent sprawl. As organizations spin up agents across different teams, frameworks, and models, it becomes hard to know which agents exist, what they can access, and whether they follow policy. Agent Manager provides a single place to deploy, observe, and govern those agents while keeping data under the organizations own control. The platform centers on governance. It gives each agent a verifiable identity with role-based access control and instant revocation, and enforces more than 40 built-in guardrails, including personal data masking, URL and content validation, and semantic prompt validation, at the organization, agent, MCP, and LLM levels, with policies mapped to the OWASP Top 10. It adds versioned lifecycle management with one-click agent suspension and a Kubernetes-native sandboxed runtime for isolation. Because it is built on OpenTelemetry, OpenAPI, and the Model Context Protocol, it works out of the box with frameworks such as LangChain, CrewAI, Amazon Bedrock Strands, and Microsoft Agent Framework, and custom implementations can hook in without rewriting existing code. Agent Manager is released under the Apache 2.0 license, so teams can self-host it with full data sovereignty or adopt a managed SaaS option. It first appeared in beta in June 2026 and reached general availability in mid-September 2026, when WSO2 added per-agent and per-environment identity controls, MCP-level governance, and the sandboxed runtime. It is a fit for enterprises that want federated management of agents across cloud, on-premise, and hybrid deployments without handing control to a closed platform.