What is an MCP Proxy Server?

An MCP Proxy Server aggregates and serves multiple MCP resource servers through a single HTTP server, simplifying integration for AI Agents.

Why do I need an MCP Proxy Server for my AI Agent?

It simplifies integration, enhances scalability, improves security, facilitates monitoring, and enables real-time updates for your AI Agent.

What types of MCP clients are supported?

The MCP Proxy Server supports stdio, sse, and streamable-http client types, providing flexibility for different integration scenarios.

How do I configure the MCP Proxy Server?

The server is configured using a JSON file, allowing you to define the proxy settings, MCP server connections, and authentication tokens.

What is toolFilter and how does it work?

toolFilter allows you to control which tools are exposed by each MCP server, enhancing security by blocking specific tool names.

What is the role of authTokens in MCP proxy?

authTokens are authentication tokens for the client. The Authorization header will be checked against this list. You can configure them at the proxy level for default access or at the individual MCP server level for specific access control.

Where can I find pre-built MCP server integrations?

The UBOS Asset Marketplace provides a curated selection of tools and resources, including pre-built MCP server integrations.

How does UBOS enhance AI Agent development?

UBOS offers orchestration, data connection, customization, and multi-agent system support, streamlining the entire AI Agent lifecycle.

How do I access the server?

You can access the server at http(s)://{baseURL}/{clientName}/sse. (e.g., https://mcp.example.com/fetch/sse, based on the example configuration)

Can I deploy MCP Proxy with Docker?

Yes, you can deploy MCP Proxy with Docker. The Docker image supports two MCP calling methods by default: npx and uvx.