GenPark Adaptive Batch Request Debouncer

Local MCP server that debounces queries and coalesces them into micro-batches to improve LLM throughput.

What you can do

• Debounces incoming queries using an adaptive sliding window.
• Coalesces related requests into micro-batches before they are processed.
• Aims to raise LLM throughput by reducing the number of separate calls.
• No individual tools are listed, so the exact operations the assistant can invoke are not documented.

When to use it

Worth considering when many similar or bursty LLM requests need to be grouped and timed before processing, and you want that batching handled locally on your machine.

Requirements

Runs as a local stdio process on the user's machine, distributed as an MCPB package for desktop clients. No authentication, environment variables or headers are declared.