AIRIS MCP Gateway
Free基于 Docker 的 MCP 多路复用器，通过 7 个元工具聚合 60+ 工具，上下文 token 减少 97%。一键启动，按需自动启用服务器。
FreeFree tier
About AIRIS MCP Gateway
AIRIS MCP Gateway is an open-source, observability-first gateway for the Model Context Protocol (MCP). It acts as a universal hub that allows AI clients—such as Claude Code, Codex, Gemini CLI, Cursor, and Windsurf—to connect to multiple backend MCP servers through a single, shared endpoint. The gateway aggregates over 60 individual tools into seven meta tools, reducing context token usage by up to 97%. It supports both Streamable HTTP and SSE protocols, features a simple two-step installation (gateway + client registration), and is designed for teams that need to standardize MCP access across projects.
Key Features
Observability-first design for monitoring MCP tool usage
Aggregates 60+ tools into 7 meta tools, reducing context token usage by 97%
One-click Docker launch with auto-enable of MCP servers
Supports both Streamable HTTP (Claude Code, Codex) and SSE (Gemini CLI, Cursor, Windsurf)
Single user-scoped registration shared across all projects
Easy install via curl script or Docker Compose from source
Open source under Apache 2.0 license
Pros & Cons
Pros
- Open source and free to use
- Reduces token consumption significantly (97% claimed)
- Supports multiple popular AI clients out of the box
- Simple two-step install process
- Shared registration simplifies multi-project setups
- Observability-first with health endpoint and logs
Cons
- Requires Docker for deployment
- Needs manual API key configuration (or Doppler integration)
- Limited to MCP protocol; not a generic API gateway
- Dependence on backend MCP servers being operational
Best For
Developers using Claude Code, Codex, or Gemini CLI who want to manage many MCP tools without per-client configurationTeams standardizing MCP access across projects with a unified gatewayOrganizations centralizing observability and monitoring of AI tool callsRapid prototyping with auto-enabling MCP servers on demand
FAQ
How do I install AIRIS MCP Gateway?
Two options: (A) Quick install via curl -fsSL https://raw.githubusercontent.com/agiletec-inc/airis-mcp-gateway/main/install.sh | bash (no source checkout needed); (B) From source by cloning the repo and running docker compose up -d after filling .env with API keys.
How do I connect an AI client?
Register the gateway as a user-scoped MCP server. For example: Claude Code uses 'claude mcp add --transport http --scope user airis-gateway http://localhost:9400/mcp/'. SSE clients (Gemini CLI, Cursor, Windsurf) use http://localhost:9400/sse.
What protocols does the gateway support?
Streamable HTTP (for Claude Code and Codex) at http://localhost:9400/mcp/ and SSE (for Gemini CLI, Cursor, Windsurf) at http://localhost:9400/sse.
Can I share the gateway across multiple projects?
Yes, the gateway is registered once per user and is shared across all projects—no separate plugin install needed.