gate22
Open-source MCP gateway and control plane for teams to govern which tools agents can use, what they can do, and how it’s audited—across agentic IDEs like Cursor, or other agents and AI tools.
claude mcp add --transport stdio aipotheosis-labs-gate22 node backend/server.js \ --env PORT="3000" \ --env LOG_LEVEL="info" \ --env CONFIG_PATH="config/gate22.json"
How to use
Gate22 acts as an open-source MCP gateway and control plane that governs which tools agents can access, how they are used, and how their activity is audited across integrated MCP endpoints. It provides a unified surface with two core functions—search and execute—so developers and automation agents can discover available tools and invoke them in a governed, auditable manner. The platform supports per-MCP allow-lists, admin-configured credential modes (organization-wide or per-user), and bundle creation so teams can assemble bespoke toolsets from permitted MCPs, then expose them through a single endpoint for their IDEs and agents. This enables scalable governance for tool usage across large teams and multi-MCP environments, while maintaining lean runtime context for the agent workflows.
To use Gate22, start the backend gateway server and point your client tooling at its unified MCP endpoint. Admins can configure which MCPs and which actions are allowed, and developers or CI agents can perform searches to enumerate available functions and then execute only the permitted actions. The system provides auditability for every call, plus dynamic resolution of tools at runtime to keep the working context compact. As a result, you can orchestrate complex toolchains across Notion, BI tools, internal APIs, or any MCP-enabled service, while preserving strict governance and visibility.
How to install
Prerequisites:
- Node.js (recommended LTS) installed on your machine or server
- Git to clone repositories
- Basic understanding of MCP endpoints and configuration files
Step 1: Install Node.js (if not already installed)
- Windows: download and run the Node.js installer from https://nodejs.org/
- macOS/Linux: use your package manager, e.g., brew install node or sudo apt-get install nodejs npm
Step 2: Clone the Gate22 MCP gateway repository
- git clone https://github.com/aipotheosis-labs/mcp-gateway.git
- cd mcp-gateway
Step 3: Install dependencies
- npm install
Step 4: Configure the server
- Copy the example config and modify as needed:
- cp config/example-gate22.json config/gate22.json
- edit config/gate22.json to include your MCP endpoints, allow-lists, and credentials modes
Step 5: Run the gateway server
- npm run start
- Or: node backend/server.js (if you run directly from the source)
Step 6: Verify
- Access the MCP gateway endpoint (default http://localhost:3000) and test the search and execute capabilities against permitted MCPs using your IDE or HTTP client.
Optional: Docker
- If a Docker image is provided for Gate22, you can run:
- docker run -p 3000:3000 -v /path/to/gate22/config.json:/app/config.json your-docker-registry/gate22:latest
Additional notes
Tips and common issues:
- Ensure the environment variables in your config.json align with your deployment (e.g., correct PORT, credentials sources, and allowed MCP endpoints).
- If you encounter authentication/permission issues, verify the admin-configured credential mode (org-shared vs per-user) and the per-MCP allow-list entries.
- The two-function design (search and execute) resolves tools at runtime; keep your bundles lean and monitor tool-change diffs to track updates.
- When running in production, consider wiring Gate22 behind a reverse proxy and enabling TLS. Enable audit logging and export logs to your SIEM for compliance.
- Check MCP endpoint compatibility and ensure each MCP’s authentication method is supported by Gate22.
- If using Docker, ensure proper volume mounts for configuration and persistence of audit logs.
Related MCP Servers
mcp-agent
Build effective agents using Model Context Protocol and simple workflow patterns
klavis
Klavis AI (YC X25): MCP integration platforms that let AI agents use tools reliably at any scale
aci
ACI.dev is the open source tool-calling platform that hooks up 600+ tools into any agentic IDE or custom AI agent through direct function calling or a unified MCP server. The birthplace of VibeOps.
mcp
🤖 Taskade MCP · Official MCP server and OpenAPI to MCP codegen. Build AI agent tools from any OpenAPI API and connect to Claude, Cursor, and more.
neurolink
Universal AI Development Platform with MCP server integration, multi-provider support, and professional CLI. Build, test, and deploy AI applications with multiple ai providers.
mcp -js
MCP server that exposes YepCode processes as callable tools for AI platforms. Securely connect AI assistants to your YepCode workflows, APIs, and automations.