dpdispatcher-mcp-server

dpdispatcher-mcp-server

1
research_and_datacloud_platforms

The Dispatcher MCP Server provides a way to manage computational jobs using the dpdispatcher library. It enables language models and MCP clients to submit, track, and manage jobs on local or HPC systems, with support for interactive configuration and stdio transport.

Dispatcher MCP Server

An MCP server that functions as a wrapper around the dpdispatcher library, offering tools to manage computational jobs on local or HPC systems.

Features

  • Exposes dpdispatcher functionality via MCP.
  • Submit, query, cancel jobs, and fetch results.
  • Includes resources for interactive job configuration.
  • Supports stdio transport for local integration.

Setup

  • Requires Python 3.x and the dpdispatcher, mcp, and anyio libraries.
  • Ensure necessary files are placed in a directory and configure dpdispatcher for remote HPC submissions.

Running the Server

  • Run fast_server.py to start the server, listening via stdio.

Usage

  • Interact using an MCP client to directly call tools or through interactive configuration processes.