What is Mcp Inception

mcp-inception is a TypeScript-based MCP server that allows one MCP client to call another MCP client. It facilitates task delegation, offloading of context windows, and enables parallel and map-reduce execution of tasks.

Use cases

Use cases include querying multiple LLMs for information simultaneously, processing large datasets in parallel, and combining results from different sources to generate comprehensive outputs.

How to use

To use mcp-inception, set up the server and client as per the README instructions. Utilize the provided tools like execute_mcp_client, execute_parallel_mcp_client, and execute_map_reduce_mcp_client to perform tasks and queries efficiently.

Key features

Key features include the ability to offload context windows, delegate tasks, execute queries in parallel, and process multiple items with map-reduce functionality. It also integrates with mcp-client-cli for enhanced performance.

Where to use

mcp-inception can be used in various fields such as natural language processing, data analysis, and any application requiring efficient task delegation and processing across multiple models.

Disclaimer

Ok this is a difficult one. Will take some setting up unfortunately.
However, if you manage to make this more straightforward, please send me PR’s.

mcp-inception MCP Server

Call another mcp client from your mcp client. Delegate tasks, offload context windows. An agent for your agent!

This is a TypeScript-based MCP server that implements a simple LLM query system.

• MCP Server and Client in one
• Made with use of mcp-client-cli
• Offload context windows
• Delegate tasks
• Parallel and map-reduce execution of tasks

Features

Tools

• execute_mcp_client - Ask a question to a separate LLM, ignore all the intermediate steps it takes when querying it’s tools, and return the output.
  • Takes question as required parameters
  • Returns answer, ignoring all the intermediate context
• execute_parallel_mcp_client - Takes a list of inputs and a main prompt, and executes the prompt in parallel for each string in the input.
  E.G. get the time of 6 major cities right now - London, Paris, Tokyo, Rio, New York, Sidney.
  • takes main prompt “What is the time in this city?”
  • takes list of inputs, London Paris etc
  • runs the prompt in parallel for each input
  • note: wait for this before using this feature
• execute_map_reduce_mcp_client - Process multiple items in parallel and then sequentially reduce the results to a single output.
  • Takes mapPrompt with {item} placeholder for individual item processing
  • Takes reducePrompt with {accumulator} and {result} placeholders for combining results
  • Takes list of items to process
  • Optional initialValue for the accumulator
  • Processes items in parallel, then sequentially reduces results
  • Example use case: Analyze multiple documents, then synthesize key insights from all documents into a summary

Development

Dependencies:

• Install mcp-client-cli
  • Also install the config file, and the mcp servers it needs in ~/.llm/config.json
• create a bash file somewhere that activates the venv and executes the llm executable

#!/bin/bash
source ./venv/bin/activate
llm --no-confirmations

install package

Install dependencies:

npm install

Build the server:

npm run build

For development with auto-rebuild:

npm run watch

Installation

To use with Claude Desktop, add the server config:

On MacOS: ~/Library/Application Support/Claude/claude_desktop_config.json
On Windows: %APPDATA%/Claude/claude_desktop_config.json

Debugging

Since MCP servers communicate over stdio, debugging can be challenging. We recommend using the MCP Inspector, which is available as a package script:

npm run inspector

The Inspector will provide a URL to access debugging tools in your browser.