"Vibe-Trading: Your Personal Trading Agent"
Terminal task manager with CLI, TUI, and MCP
Taskdog gives you a local task manager you can use from the terminal or through a REST API. It keeps data in SQLite and includes scheduling, time tracking, dependency checks, search, and Gantt-style views. It also ships an MCP server for Claude Desktop.
Builders who want a local terminal task manager with API and MCP access.
You can track and reschedule work from the terminal without juggling separate tools.
What it does
Multiple interfaces
Use the same task data from the CLI, the full-screen TUI, or the REST API.
Schedule optimization
Run several scheduling algorithms, including greedy, genetic, and Monte Carlo approaches.
Search and filter
Use fzf-style search, progressive filter chains, and multi-field sorting.
Time tracking
Compare planned time with actual time and track task progress automatically.
Task dependencies
Link tasks and detect circular dependencies.
Markdown notes
Attach notes with editor integration and Rich rendering.
MCP integration
Connect Claude Desktop through a Model Context Protocol server.
How to get it
- 1Try taskdog with ~50 sample tasks. No installation required — just Docker
docker run --rm -it ghcr.io/kohei-wada/taskdog:demo
- 2The TUI works inside the container, but some keybindings (e.g., Ctrl+P for command…
docker run --rm -d -p 8000:8000 --name taskdog-demo ghcr.io/kohei-wada/taskdog:demo # Wait for the server and demo data to be ready (~15s) docker logs -f taskdog-demo 2>&1 | grep -m1 "Server ready" uvx --from taskdog-ui taskdog tui
- 3Run
git clone https://github.com/Kohei-Wada/taskdog.git cd taskdog make install
- 4Run
yay -S taskdog
- 5Run
pip install taskdog-ui[server]
- 6Run
taskdog add "My first task" --priority 10 taskdog list taskdog gantt taskdog tui
README
A task management system with CLI/TUI interfaces and REST API server, featuring time tracking, schedule optimization, and beautiful terminal output.
Designed for individual use. Stores tasks locally in an SQLite database.
https://github.com/user-attachments/assets/2c0de3ec-fa3d-4f41-ae01-acbff04931e3
Screenshots
TUI (Textual)

Gantt Chart (CLI)

Try It Out
Try taskdog with ~50 sample tasks. No installation required — just Docker:
docker run --rm -it ghcr.io/kohei-wada/taskdog:demo
The TUI works inside the container, but some keybindings (e.g., Ctrl+P for command palette) may conflict with Docker's key sequences. For the best experience, run the server in a container and connect from your host:
docker run --rm -d -p 8000:8000 --name taskdog-demo ghcr.io/kohei-wada/taskdog:demo
# Wait for the server and demo data to be ready (~15s)
docker logs -f taskdog-demo 2>&1 | grep -m1 "Server ready"
uvx --from taskdog-ui taskdog tui
uvxcomes with uv. It runs the command in a temporary environment without installing anything.
Installation
Requirements: Python 3.12+, uv
Supported Platforms: Linux, macOS
Experimental: Windows (WSL2 recommended; native support is still being hardened)
Recommended (with systemd/launchd service)
git clone https://github.com/Kohei-Wada/taskdog.git
cd taskdog
make install
This installs the CLI/TUI and server, and sets up a systemd (Linux) or launchd (macOS) service so the server starts automatically.
Arch Linux (AUR)
yay -S taskdog
Installs the CLI/TUI, server, and MCP binaries plus the systemd user service
(enable per-user with systemctl --user enable --now taskdog-server). See
contrib/aur/ for packaging details.
From PyPI
pip install taskdog-ui[server]
You'll need to manage the server process yourself (e.g., taskdog-server &).
Windows Users
- WSL2 is recommended and follows the same setup flow as Linux.
- Native Windows support is experimental. By default, data is stored under
%LOCALAPPDATA%\taskdogand configuration under%APPDATA%\taskdog. - The editor integration checks
%EDITOR%first, then falls back tocode,notepad, andvim.
Usage
taskdog add "My first task" --priority 10
taskdog list
taskdog gantt
taskdog tui
For complete setup including API key configuration, see Quick Start Guide.
Features
- Multiple Interfaces: CLI, full-screen TUI, and REST API
- Schedule Optimization: 9 algorithms (greedy, genetic, monte carlo, etc.)
- Search & Filter: fzf-style queries, progressive filter chains, multi-field sort
- Time Tracking: Automatic tracking with planned vs actual comparison
- Gantt Chart: Visual timeline with workload analysis
- Task Dependencies: With circular dependency detection
- Markdown Notes: Editor integration with Rich rendering
- Audit Logging: Track all task operations
- MCP Integration: Claude Desktop support via Model Context Protocol
Search, filter & sort
Maximize the task list, sort it from the command palette, then narrow it down with / search, Ctrl+U to re-search, and Ctrl+R to build a progressive filter chain (substring + status/tag exclusion):
https://github.com/user-attachments/assets/dcb44390-7b10-49a0-bdfc-01a03d7751f9
Documentation
- Quick Start Guide - Step-by-step setup
- CLI Commands Reference - Complete command documentation
- API Reference - REST API endpoints and examples
- Configuration Guide - All configuration options
- Design Philosophy - Why Taskdog works this way
- Deployment Guide - Docker, systemd, launchd
Architecture
UV workspace monorepo with five packages:
| Package | Description | PyPI |
|---|---|---|
| taskdog-core | Core business logic and SQLite persistence | |
| taskdog-client | HTTP API client library | |
| taskdog-server | FastAPI REST API server | |
| taskdog-ui | CLI and TUI interfaces | |
| taskdog-mcp | MCP server for Claude Desktop |
Contributing
Contributions are welcome! See CONTRIBUTING.md for guidelines.
License
MIT License - see LICENSE for details.
Files in the repo
- .github
- contrib
- docs
- examples
- packages
- scripts
- tests
- vhs
- .codespellrc
- .dockerignore
- .env.example
- .gitignore
- .markdownlint.yaml
- .pre-commit-config.yaml
- .python-version
- CLAUDE.md
- CONTRIBUTING.md
- docker-compose.yaml
- Dockerfile
- Dockerfile.demo
- flake.lock
- flake.nix
- LICENSE
- Makefile
- mkdocs.yml
- pyproject.toml
- README.md
- uv.lock
Discussion (0)
Ask about usage, or say what you built with itSign in to join the discussion.
No comments yet. Be the first to say what this is good for.
More other
OpenHuman is an open source personal AI for Mac, Windows and Linux — local-first memory, agent orchestration, and deep research.

Your Personal AI Assistant; easy to install, deploy on your own machine or on the cloud; supports multiple chat apps with easily extensible capabilities.
macOS video editor built for AI
Open-source Claude Design alternative. One-click import your Claude Code / Codex API key. Prompt → prototype / slides / PDF. Multi-model (Claude, GPT, Gemini, Kimi, GLM, Ollama). BYOK, local-first, MIT.
🤯 LobeHub is your Chief Agent Operator, organizing your agents into 7×24 operations by hiring, scheduling, and reporting on your entire AI team.