An open-source proxy that enforces identity, policy, and audit controls between AI agents and enterprise systems without exposing upstream credentials.
Adapted from @DanKornasGiving an AI agent a production credential shouldn’t mean giving it the keys. Warden is a secure gateway for connecting AI agents to enterprise systems such as MCP servers, cloud services, code hosts, observability tools, databases, and SaaS. It helps you put identity, policy, and audit in the request path: the agent presents its identity, Warden evaluates the request, injects the upstream credential, and proxies the response. Key features: • Identity-based access – supports JWTs, TLS client certificates, SPIFFE SVIDs, and Kubernetes service-account tokens • Per-request policy – evaluates access when each call happens, including MCP tool calls and arguments • Credential isolation – agents can work without holding upstream secrets in their environment • Broad provider coverage – fronts MCP, LLM, cloud, code-hosting, observability, ITSM, Kubernetes, secrets, and database systems • Audit and attribution – records the identity, role, policy decision, and upstream for each request It’s open-source (Mozilla Public License 2.0 license). Link in the reply 👇
