TaskMateAI
AI/MCP TODO task management application
TaskMateAI is a simple task management application that enables AI to autonomously manage and execute tasks, and can be operated through MCP (Model Context Protocol).
README available here
Features
- Creating and managing tasks through MCP
- Subtask Support
- Priority-based task handling
- Task progress management and reporting function
- Add notes feature
- Data persistence via JSON files
- Task management for multiple AIs by agent ID
- Organizing tasks by project
install
Prerequisites
- Python 3.12 or higher
- uv (Python package manager)
- WSL (Windows Subsystem for Linux) *For Windows environments
Installation Instructions
- Clone or download the repository:
git clone https://github.com/YourUsername/TaskMateAI.git
cd TaskMateAI
- Install the required packages:
uv install -r requirements.txt
How to use
Application launch
In the WSL environment you can run your application like this:
cd /path/to/TaskMateAI/src/TaskMateAI
uv run TaskMateAI
MCP Configuration
Example of configuration for use with MCP:
{
"mcpServers": {
"TodoApplication": {
"command": "uv",
"args": [
"--directory",
"/絶対パス/TaskMateAI",
"run",
"TaskMateAI"
],
"env": {},
"alwaysAllow": [
"get_tasks", "get_next_task", "create_task", "update_progress",
"complete_task", "add_subtask", "update_subtask", "add_note",
"list_agents", "list_projects"
],
"defaultArguments": {
"agent_id": "agent_123",
"project_name": ""
}
}
}
}
{
"mcpServers": {
"TodoApplication": {
"command": "wsl.exe",
"args": [
"-e",
"bash",
"-c",
"cd /絶対パス/TaskMateAI && /home/ユーザー/.local/bin/uv run TaskMateAI"
],
"env": {},
"alwaysAllow": [
"get_tasks", "get_next_task", "create_task", "update_progress",
"complete_task", "add_subtask", "update_subtask", "add_note",
"list_agents", "list_projects"
],
"defaultArguments": {
"agent_id": "agent_123",
"project_name": ""
}
}
}
}
TaskMateAI provides the following MCP tools:
- get_tasks - Get a list of tasks (can be filtered by status and priority)
- get_next_task - Get the next high priority task (automatically updates to in progress status)
- create_task - Create a new task (with subtasks)
- update_progress - Updates the progress of a task
- complete_task - Mark a task as complete
- add_subtask - Add a subtask to an existing task
- update_subtask - Update the status of a subtask
- add_note - Add a note to a task
- list_agents - Get a list of available agent IDs
- list_projects - Get a list of projects related to a specific agent
Tasks are managed using the following structure:
{
"id": 1,
"title": "タスクのタイトル",
"description": "タスクの詳細な説明",
"priority": 3,
"status": "todo", // "todo", "in_progress", "done" のいずれか
"progress": 0, // 0-100 の進捗率
"subtasks": [
{
"id": 1,
"description": "サブタスクの説明",
"status": "todo" // "todo", "in_progress", "done" のいずれか
}
],
"notes": [
{
"id": 1,
"content": "ノートの内容",
"timestamp": "2025-02-28T09:22:53.532808"
}
]
}
Data Storage
Task data is stored in a hierarchical structure:
output/
├── tasks.json # デフォルトのタスクファイル
├── agent1/
│ ├── tasks.json # agent1のタスクファイル
│ ├── project1/
│ │ └── tasks.json # agent1のproject1のタスクファイル
│ └── project2/
│ └── tasks.json # agent1のproject2のタスクファイル
└── agent2/
├── tasks.json # agent2のタスクファイル
└── projectA/
└── tasks.json # agent2のprojectAのタスクファイル
Each task file is automatically generated and updated when the application is run.
Managing agents and projects
To manage tasks for a specific agent or project, you can:
- Specify a default agent in your MCP settings : By specifying
agent_id in defaultArguments , it will be used automatically in all requests. - Specify projects in AI conversations : You can specify projects in the conversation, such as "Add a new task to project X."
- Directly specified by AI : You can include
agent_id and project_name in the request parameters.
Project Structure
TaskMateAI/
├── src/
│ └── TaskMateAI/
│ ├── __init__.py # パッケージ初期化
│ └── __main__.py # メインアプリケーションコード
├── output/ # データ保存ディレクトリ
│ └── tasks.json # タスクデータ (自動生成)
├── tests/ # テストコード
│ ├── unit/ # ユニットテスト
│ └── integration/ # 統合テスト
├── requirements.txt # 依存パッケージリスト
└── README.md # このファイル
test
TaskMateAI provides a comprehensive test suite to ensure functionality reliability.
Test Configuration
The tests are organized in the following directory structure:
tests/
├── __init__.py # テストパッケージの初期化
├── conftest.py # テスト用フィクスチャの定義
├── unit/ # ユニットテスト
│ ├── __init__.py
│ ├── test_task_utils.py # タスク関連ユーティリティのテスト
│ ├── test_mcp_tools.py # MCPツール機能のテスト
│ └── test_agent_projects.py # エージェントとプロジェクト管理のテスト
└── integration/ # 統合テスト
└── __init__.py
Test types
- Unit testing : Ensures that individual components of an application function correctly
test_task_utils.py : Tests basic functions such as reading and writing tasks, and generating IDs.test_mcp_tools.py : Tests the functionality of MCP tools (creating, updating, completing tasks, etc.)test_agent_projects.py : Tests agent ID and project management functionality
- Integration testing : Ensure that multiple components work together correctly (future expansion planned)
How to run the test
You can run the test using the following command:
- Run all tests:
cd /path/to/TaskMateAI
uv run python -m pytest -xvs
- Run a specific test file:
uv run python -m pytest -xvs tests/unit/test_task_utils.py
- Run a specific test class:
uv run python -m pytest -xvs tests/unit/test_mcp_tools.py::TestMCPTools
- Run a specific test function:
uv run python -m pytest -xvs tests/unit/test_task_utils.py::TestTaskUtils::test_read_tasks_with_data
Explanation of test arguments:
-x : Stops testing when an error occurs-v : Display verbose output-s : Display standard output during testing
Items to be fixed
- Implementing the task template function
- Building a dependency management system between tasks
- Addition of schedule function
- Introducing a tag-based task classification system
- Implementing milestone management function
license
MIT
author
New AI Tees