Persistproc

This MCP server provides AI assistants with persistent process management capabilities for multi-agent development workflows, built by Steve Landey using Python with FastMCP and rich console output to enable natural language control over long-running development processes like servers, build watchers, and background tasks. The implementation offers unified process control through start/stop/restart operations with custom labeling, real-time output streaming and capture, process discovery and filtering, and robust cleanup handling, featuring both command-line interface and MCP server modes with JSON/text output formats and comprehensive logging to isolated data directories. Built with subprocess management using process groups for clean termination, file-based log capture with streaming support, environment variable injection, and extensive test coverage using pytest with real process lifecycle testing, it serves development teams needing conversational access to background process orchestration, DevOps engineers requiring AI-driven service management, and multi-agent workflows where natural language interfaces enhance development productivity without manual process juggling across terminal sessions.

Source

Repository: https://github.com/irskep/persistproc