Agentgateway
Content Lifecycle & Evolution State
Overview
Agentgateway is an open source proxy built on AI-native protocols that provides drop-in security, observability, and governance for agent-to-LLM, agent-to-tool, and agent-to-agent communication.
Key Features
LLM Gateway: Route traffic to major LLM providers through a unified OpenAI-compatible API with budget controls, load balancing, and failover.
MCP Gateway: Connect LLMs to tools and external data sources via MCP with tool federation, OpenAPI integration, and OAuth authentication.
A2A Gateway: Secure agent-to-agent communication using A2A with capability discovery and task collaboration.
Inference Routing: Intelligent routing to self-hosted models using Kubernetes Inference Gateway extensions based on GPU utilization, KV cache, and queue depth.
Guardrails: Multi-layered content filtering with regex, OpenAI moderation, AWS Bedrock Guardrails, Google Model Armor, and custom webhooks.
Security & Observability: Auth (JWT, API keys, OAuth), fine-grained RBAC with CEL policy engine, rate limiting, TLS, and OpenTelemetry metrics/logs/tracing.
Use Cases
Routing and securing traffic between AI agents, LLMs, and external tools/services. Providing a unified gateway for agent-to-agent and agent-to-tool communications with embedded security, routing logic, and observability out-of-the-box.
Getting Started
Check the standalone or Kubernetes quickstart guides on agentgateway.dev to deploy the proxy using a flat YAML config or the built-in Kubernetes controller.
Recent Updates
Release v1.5.0 introduces numerous enhancements including support for forward proxy authentication, limit on authorized models per API key, connecting tunneling through dynamic proxy backends, and API-key scoped budgets for LLM traffic. It also includes UI fixes and various security refinements.