Connect Secure Observe Connect Agentic Workflows
Agent Gateway is an open source data plane built on AI-native protocols (A2A & MCP) to connect, secure, and observe agent-to-agent and agent-to-tool communication across any framework and environment.
Contributing Companies
Solo.io
Microsoft
Apple
Alibaba
Adobe
AWS
Cisco
Salesforce
Huawei
Amdocs
What is Agent Gateway?
Agent Gateway is a next-generation proxy designed for the agentic AI ecosystem. It provides drop-in security, observability, and governance for agent-to-agent (A2A) and agent-to-tool (MCP) communication.
Built to tackle enterprise challenges, Agent Gateway enables teams to connect, secure, and audit all AI agent communications from a single control point.
Protocol Native
Built on MCP and A2A protocols for seamless agent connectivity
Security First
RBAC, JWT authentication, TLS, and CEL-based access policies
High Performance
Written in Rust, designed for any scale deployment
Features
Everything you need to connect, secure, and observe your AI infrastructure
Seamlessly switch between providers without changing your application code.
| Provider | Chat Completions | Streaming |
|---|---|---|
| OpenAI / Azure OpenAI | ✓ | ✓ |
| Anthropic | ✓ | ✓ |
| Google Gemini | ✓ | ✓ |
| Google Vertex AI | ✓ | ✓ |
| Amazon Bedrock | ✓ | ✓ |
OpenAI-compatible providers
Route to any provider that supports the OpenAI API format:
Cohere, Mistral, Groq, Together AI, Fireworks, Ollama, LM Studio, vLLM, llama.cpp, and any custom endpoint with /v1/chat/completions
Running your own models on GPU infrastructure? Agentgateway implements the Kubernetes Inference Gateway extensions for intelligent routing to local LLM workloads.
Route based on:
- •GPU & KV cache utilization — Send requests to the least-loaded model
- •Prompt criticality — Prioritize high-priority requests
- •LoRA adapters — Route to models with specific fine-tuned adapters
- •Work queue depth — Avoid overloaded inference servers
Connect LLMs to tools and external data sources using the Model Context Protocol (MCP).
- •Tool federation — Aggregate multiple MCP servers behind a single endpoint
- •Protocol support — stdio, HTTP/SSE, and Streamable HTTP transports
- •OpenAPI integration — Expose existing REST APIs as MCP-native tools
- •Authentication & authorization — Built-in MCP auth spec compliance with OAuth providers (Auth0, Keycloak)
Enable secure communication between AI agents using the Agent-to-Agent (A2A) protocol. Agents can:
- •Discover each other's capabilities
- •Negotiate interaction modalities (text, forms, media)
- •Collaborate on long-running tasks
- •Operate without exposing internal state or tools
Authentication
JWT, API keys, basic auth, MCP auth spec
Authorization
Fine-grained RBAC with Cedar policy engine
Traffic Policies
Rate limiting, CORS, TLS, external authz
Observability
Built-in OpenTelemetry metrics, logs, and distributed tracing
Getting Started
View all docscurl https://raw.githubusercontent.com/agentgateway/agentgateway/refs/heads/main/common/scripts/get-agentgateway | bashcurl -sL https://raw.githubusercontent.com/agentgateway/agentgateway/main/examples/basic/config.yaml -o config.yamlagentgateway -f config.yamlUse Cases
View all tutorialsMCP Federation
Expose a single MCP endpoint that federates tools from multiple backend MCP servers with unified security.
MCP 4 tutorialsLLM Gateway
Route requests to OpenAI, Anthropic, Azure with built-in rate limiting and cost tracking.
LLM 3 tutorialsAgent-to-Agent
Enable secure communication between agents using the A2A protocol across different frameworks.
A2APopular Integrations
View all integrationsLLM Providers
Connect to OpenAI, Anthropic, Azure OpenAI, Amazon Bedrock, and Google Gemini.
MCP Servers
Expose and federate MCP tool servers with stdio, SSE, and streamable HTTP transports.
Kubernetes Gateway API
Deploy with kgateway for dynamic provisioning using Kubernetes Gateway API.
Observability
OpenTelemetry, Prometheus, Grafana, and Jaeger for metrics, tracing, and visualization.
AI-native connectivity for agentic applications
Tool Federation
Provide a single MCP endpoint for all the tools your agents consume, with unified security, observability, and governance for all agent-to-tool communication.
Unified Connectivity
Support for industry standard AI protocols for agent and tool connectivity including A2A and MCP with the ability to automatically expose existing REST APIs as MCP-native tools.
Developer Portal
Self-service developer portal for agent and tool developers to create, configure, discover, and debug tools and agents from a single pane of glass UI.
Join the Community
Calling all agent creators, tool providers, platform engineers, and AI enthusiasts - come build the future of AI agent connectivity.
Weekly Community Meeting
Join us for our weekly community sync to discuss features, roadmap, and community contributions.
“The future won’t be built by standalone agents, MCP servers or LLMs — it’s shaped by their interconnection and ability to work together seamlessly. To unlock their full potential, we must apply policies, ensure control and maintain clear visibility into their interactions. This is where agentgateway plays a pivotal role — bridging not only agent-to-agent (A2A) communication but also agent-to-MCP servers, filling a critical gap in the ecosystem. I look forward to seeing the project’s continued momentum within the Linux Foundation.”
Solving AI Connectivity Challenges
Agent and Tool Interoperability
Built on the leading industry protocols for agent and tool connectivity, agentgateway allows you to seamlessly integrate any agent and tool supporting A2A and MCP.
Agent Governance
Agentic applications will be composed of multiple tools and agents working together to achieve a goal, creating a fragmented landscape for security and observability. Agentgateway is a drop-in solution transparent to agents and tools to secure, govern, and audit agent and tool communications.
Tool Sprawl
Agent development will never scale treating every tool integration as a 1:1 integration. Agentgateway provides a federated MCP endpoint with a centralized registry, self-service discovery, and dynamic configuration of agents and tools.
Leveraging Existing APIs
No need to create custom MCP tool server implementations for every REST API you have today. Agentgateway automatically translates OpenAPI resources into MCP tools ready to consume from agents.