agent-pm

Agent Pm MCP Server

AgentPM helps developers turn software requirements into prioritized tasks for AI-assisted implementation. Its local MCP integration supports project brief interviews, dependency-aware task selection, task editing, and library documentation retrieval.

Server facts

agent-pm MCP Server

Project

Author
gannonh
License
MIT license

Tech

Runtime / language
node
Transports
stdio
Remote available
No
Authentication
Not specified

Activity

GitHub stars
30
Last commit
Data as of

Tags

  • Cursor
  • cline ai
  • co pilot
  • cursor ai
  • mcp server

Overview

Development work in AgentPM can begin with existing project material or an interview that produces a project brief and tasks. The brief workflow exposes separate tools to start the interview operation, check its progress, and retrieve the completed result. From there, apm_task lets a coding agent inspect individual tasks, filter work by status or priority, and choose the next task using dependency and completion information. AgentPM runs locally through MCP for developers using compatible environments such as Cursor, Augment, and VS Code Copilot.

Task editing is handled by apm_task_modify, which creates, changes, or deletes tasks and subtasks, while apm_dependencies manages their relationships. apm_complexity analyzes tasks and produces expansion recommendations and reports. apm_task_generate writes individual task files in apm-artifacts from artifacts.json. Documentation lookup follows a separate Context7 flow: resolve a package name to a library ID, then fetch that library's documentation. AgentPM also adjusts future tasks in response to finished work and maintains project documentation as implementation changes.

agent-pm setup notes

  • agent-pm records stdio only; confirm a network transport upstream before planning remote access.

Related MCP servers