execute_map_reduce_mcp_client
Process multiple items in parallel then sequentially reduce the results to a single output.
LABSCO SUMMARY
Context offloading, at the price of a second stack to maintain.
The idea is sound and unusual: a research task that would burn thousands of tokens of tool chatter runs in another process and returns a paragraph. The cost is honest in the README, which opens by saying it takes some setting up — you are configuring and maintaining a whole second MCP client before the first call works. The parallel tool is also flagged as waiting on an upstream mcp-client-cli change, so exercise it on a short list before building anything on top of it.
What it is
An MCP server that is also an MCP client. It sends a question to a separate LLM running its own tool stack, discards every intermediate step that model took, and returns just the result — so the work happens somewhere else and your context window only receives the answer.
What you get
- A question answered by a separate LLM, with all of its intermediate tool calls dropped and only the final output returned —
execute_mcp_client - One prompt run across a list of inputs at once, such as asking the same question of six cities —
execute_parallel_mcp_client - A list processed in parallel and then folded to a single result, with a
mapPromptcarrying an{item}placeholder and areducePromptcarrying{accumulator}and{result}, plus an optional starting value —execute_map_reduce_mcp_client
Requirements
The delegated model is your own: this drives mcp-client-cli, which you install separately along with its config at ~/.llm/config.json and whichever MCP servers that client should have. You also write a small shell script that activates its virtualenv and runs the llm executable, then point MCP_INCEPTION_EXECUTABLE at that script and MCP_INCEPTION_WORKING_DIR at the client's working directory. The npm package name is mcp-inception (0.1.0) and it is marked private, so build from a clone with npm install and npm run build.
Setup effort
Follow the published steps — git clone https://github.com/tanevanwifferen/mcp-inception
USE CASES
01
Keeping a long research detour out of your context
Delegate the question to the second model, let it make as many tool calls as it needs, and receive only the conclusion.
02
Asking the same question of many things at once
Give one prompt and a list of subjects — cities, repositories, customers — and have it run against each instead of looping in the main conversation.
03
Summarising a pile of documents into one answer
Map a per-document prompt across the set, then reduce the results with an accumulator prompt so the synthesis happens outside your context window.
04
Giving an assistant tools you do not want it to hold
Attach the risky or noisy servers to the delegated client instead of your main one, and reach them only through a single question-and-answer call.
3 tools
Counted once with only startup setup. Other versions, modes or optional credentials may expose a different list.
execute_mcp_client
Offload certain tasks to AI. Used for research purposes, do not use for code editing or anything code related. Only used to fetch data.
execute_parallel_mcp_client
Execute multiple AI tasks in parallel, with responses in JSON key-value pairs.
BEFORE YOU INSTALL
What to know first
Tools you get
3 tools when we connected.
Requires
- mcp-client-cli installed, with ~/.llm/config.json and the MCP servers it should use
- Node.js and npm (to build the clone)
- MCP_INCEPTION_EXECUTABLE and MCP_INCEPTION_WORKING_DIR set in the client entry
Where it runs
The documented client runs the server on your machine.
CONTEXT COST
Light Standing cost
Rated by the size of the 3 tool definitions that ride along on every turn, not by how many there are
INSTALLATION
Connect your client
Server setup
Install mcp-client-cli (https://github.com/adhikasp/mcp-client-cli), with its config file and the MCP servers it needs in ~/.llm/config.json
Create a bash file (e.g. run_llm.sh) that activates the mcp-client-cli venv and runs llm --no-confirmations: source ./venv/bin/activate then llm --no-confirmations
TERMINAL
git clone https://github.com/tanevanwifferen/mcp-inception
cd mcp-inception
npm install
npm run buildMCP Clients
Claude Desktop
CLAUDE DESKTOP · JSON
{
"mcpServers": {
"mcp-inception": {
"command": "node",
"args": [
"<YOUR_PATH_TO_MCP_INCEPTION>/build/index.js"
],
"env": {
"MCP_INCEPTION_EXECUTABLE": "<YOUR_MCP_INCEPTION_EXECUTABLE>",
"MCP_INCEPTION_WORKING_DIR": "<YOUR_MCP_INCEPTION_WORKING_DIR>"
}
}
}
}Point args at build/index.js in your clone, MCP_INCEPTION_EXECUTABLE at the bash file from the previous step, and MCP_INCEPTION_WORKING_DIR at the mcp-client-cli working directory.
