MCP-of-MCPS

What is MCP-of-MCPS?

MCP-of-MCPS is a meta-server that merges all your MCP servers into a single smart endpoint. It gives AI agents instant tool discovery, selective schema loading, and massively cheaper execution, turning chaotic multi-server setups into a fast, efficient, hallucination-free workflow.

How to use MCP-of-MCPS?

Install the package via npx @dbestai/mcp-of-mcps and add it to your Cline MCP settings file. Configure it with either an inline JSON array of servers or a config file path. Use the four provided tools in sequence: get_mcps_servers_overview for a lightweight list, semantic_search_tools to search by intent, get_tools_overview to load only needed schemas, and run_functions_code to execute workflows.

Key features of MCP-of-MCPS

• Semantic search over all tool descriptions by task intent
• Lightweight tool name listing without loading full schemas
• Selective loading of only the tool definitions you need
• Direct code execution between tools without model round-trips
• Persistent tool output schema metadata across sessions
• Progressive disclosure: load tools on-demand as needed

Use cases of MCP-of-MCPS

• Quickly discover which tools are available across many MCP servers
• Search for tools by describing your task in plain English
• Load only the schemas for tools you actually plan to use
• Execute multi-step data processing with minimal token consumption
• Preserve analyzed tool output schemas for consistent behavior

FAQ from MCP-of-MCPS

What is MCP-of-MCPS?

MCP-of-MCPS is a meta-server that merges all your individual MCP servers into a single endpoint, providing tool discovery, selective schema loading, and code-based tool orchestration.

What are the prerequisites to use MCP-of-MCPS?

You need Node.js, npm or yarn, and configured MCP servers that you want to aggregate.

How do I configure MCP-of-MCPS?

Add it to your Cline MCP settings with either an inline --config argument listing the server definitions or a --config-file argument pointing to a JSON config file.

How does MCP-of-MCPS reduce token usage?

It uses run_functions_code to keep data flows between tools as native objects in memory, returning only the final result to the model, and it loads only the needed tool schemas instead of all of them.

What tools does MCP-of-MCPS provide?

It provides four tools: semantic_search_tools (semantic search), get_mcps_servers_overview (list tool names), get_tools_overview (load specific tool schemas), and run_functions_code (execute JavaScript workflows).