MCP Manager

What is MCP Manager?

MCP Manager acts as a bridge between Large Language Models (LLMs) and MCP servers, enabling natural language interaction with remote and local APIs via supported chat completion APIs. Workspaces allow distinct configurations for different LLM endpoints and MCP servers using a single instance.

How to use MCP Manager?

Download the appropriate release binary for your operating system. Configure via a YAML file (default config.yaml; path can be overridden with MCP_MANAGER_CONFIG environment variable). Start the server with mcp-manager and send prompts via HTTP calls to the exposed API.

Key features of MCP Manager

• Integrates with Gemini and Azure OpenAI LLMs
• Enables LLM interaction with MCP servers
• Flexible workspace configuration via a single YAML file
• Supports API Key authentication (OAuth 2.0 planned)
• Exposes a simple HTTP API in OpenAI format

Use cases of MCP Manager

• Interacting with local file systems or APIs using natural language prompts
• Managing multiple MCP server connections with distinct LLM endpoints
• Querying or modifying systems via an LLM’s tool‑calling capabilities

FAQ from MCP Manager

Which LLM providers are supported?

Gemini and Azure OpenAI are currently supported. Claude and OpenAI integration are planned.

How does authentication work?

Only API Key authentication is currently supported, configured within the model settings. OAuth 2.0 support is planned.

Where is the configuration stored and how can I customise it?

Configuration is managed through a YAML file, defaulting to config.yaml in the runtime directory. The path can be overridden by setting the MCP_MANAGER_CONFIG environment variable.

Can I connect to remote MCP servers?

Remote MCP server connections are not yet supported; only local MCP servers are currently available. Remote support is planned.

What transport protocol does MCP Manager use?

It exposes a simple HTTP API that accepts prompts in the OpenAI format.