Agentgateway

Agentgateway is an open source gateway that handles traditional application traffic and AI native protocols in a single data plane. One binary routes, secures, observes and governs HTTP, gRPC and TCP services, LLM provider traffic, Model Context Protocol tool calls and agent to agent communication, so platform teams do not have to stitch together separate gateways for each. It is aimed at platform and infrastructure teams running agents in production, with deployment guides for both standalone use and Kubernetes. The LLM gateway exposes an OpenAI compatible API in front of OpenAI, Anthropic, Bedrock, Gemini, Vertex and open source Llama deployments, with credential handling, failover, token budgets, semantic caching and prompt redaction. Inference routing picks replicas for self hosted models on vLLM, TGI or Triton based on latency, cost and model. The MCP gateway treats MCP servers like microservices, with discovery, versioning, signed and scoped tool calls and an audit trail of what agents invoked. The A2A gateway routes calls between LangChain, CrewAI, ADK and custom runtimes with identity and tracing on each hop. TLS, mTLS and OIDC are built in, and an admin UI includes an LLM playground for testing routes. The project itself is free and open source, with no pricing page. Enterprise distributions and commercial support are available from partners; Solo.io offers Solo Enterprise for agentgateway and a production support service with round the clock coverage, both priced on request. Agentgateway was originally created by Solo.io and has joined the Agentic AI Foundation, with contributions from companies including Microsoft, AWS, Cisco and Salesforce.