mcp-batchit

MCP BatchIt exposes a single tool called batch_execute that lets you consolidate multiple MCP tool calls into one request. This reduces token usage and network chatter by executing sub-operations against a target MCP server in parallel, up to a configurable maxConcurrent limit. BatchIt handles spawning or connecting to the downstream MCP server behind the scenes and returns a single consolidated JSON result for the whole batch. It also provides controls such as stopOnError to halt remaining ops if one fails, and timeoutMs to bound each sub-op. You interact with BatchIt by sending a batch_execute request to the server’s tools/call endpoint (name=batch_execute). The server expects a structure with targetServer information, a list of operations (each specifying a tool and arguments), and options for parallelism and error handling. Typical usage includes batching filesystem-like tasks, API calls to other MCP servers, or any sequence of MCP tool invocations that can run in parallel.

Prerequisites:

• Node.js (v14+ recommended) and npm installed
• Git installed

Installation steps:

1. Clone the repository git clone https://github.com/ryanjoachim/mcp-batchit.git

2. Navigate to the project and install dependencies cd mcp-batchit npm install

3. Build the project (if a build step is defined) npm run build

4. Start the MCP BatchIt server npm start

5. Confirm the server is running (it should output a ready message and listen on STDIO for MCP tool calls).

Tips and notes:

• BatchIt’s design intentionally does not pass data between sub-ops within the same batch Execute. If a later sub-op depends on earlier results, chain multiple batch_execute calls instead.
• You can tune batch behavior with options like maxConcurrent, timeoutMs, and stopOnError. A higher maxConcurrent increases parallelism but may put more load on the downstream MCP server.
• Typical downstream targets are filesystem servers or other MCP services. BatchIt will spawn or connect to the chosen downstream server and execute all sub-ops in parallel up to the configured limits.
• Ensure the targetServer configuration for each operation correctly identifies the downstream MCP server type and transport, otherwise sub-ops may fail with tool not found or transport errors.