mcp_planning

pcingola/mcp_planning

3.2

If you are the rightful owner of mcp_planning and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to henry@mcphub.com.

An MCP server for planning Agentic tasks, similar to a 'to do' list.

MCP Planning Server

An MCP server for planning Agentic tasks, similar to a "to do" list.

Overview

This project implements a Model Context Protocol (MCP) server that provides tools for managing task planning. It allows users to create, update, and track tasks in a hierarchical structure.

Features

  • Create and manage tasks with subtasks
  • Track task state (pending, in-progress, completed, failed)
  • Persistent storage of tasks in JSON format
  • User and session isolation for task data
  • MCP tools for task management

Installation

  1. Clone the repository
  2. Install dependencies using uv:
uv pip install -e .

Usage

Running the server

./scripts/run.sh

This will start the MCP server on the configured host and port (default: 127.0.0.1:9000).

MCP Tools

The server provides the following MCP tools:

  • add_task: Add a new task
  • get_tasks: Get all tasks, optionally filtered by state
  • get_task: Get a specific task by ID
  • update_task_state: Update a task's state
  • delete_task: Delete a task
  • add_subtask: Add a subtask to an existing task

MCP Resources

The server provides the following MCP resources:

  • tasks://{user_id}/{session_id}: Get the task list for a specific user and session
  • tasks://sessions/{user_id}: Get all session IDs for a specific user
  • tasks://users: Get all user IDs that have stored task lists

Development

Running tests

./scripts/test.sh

Running linting

./scripts/lint.sh

License