qtask: a task tracker my AI coding agents share, in one Python file
I run several AI coding agents at once on the same project. One fixes a backend bug, one checks a screen in the browser, one writes tests. For a while I tracked their work in chat and in notes, and the same problems kept coming back: Two agents picked up the same task and changed the same files. An agent asked me something in the middle of a long session, I missed it, and it went ahead on a…
qtask is a task management tool for AI coding agents, developed by Abubakr Alsheikh. It is a single Python file that uses only the standard library and SQLite database to track tasks for multiple agents working on the same project. The tool addresses common issues such as agents working on the same task simultaneously, agents asking questions mid-session without tracking, agents claiming tasks without evidence, and agents not leaving any trace of their work.
qtask is implemented as a one-file Python program, with no external dependencies. It consists of around 2,000 lines of code and includes features such as task creation, claiming, asking questions, answering questions, marking tasks as done, and adding comments. The tasks are stored in a SQLite database on the user's machine, and the project information is derived from the git remote of the current working directory.
This ensures that all agents and the human collaborator share the same task list, regardless of how many copies of the project are cloned or how many worktrees exist.
The tool's workflow involves claiming a task before working on it, asking questions that are parked in a "needs_confirm" status, answering questions using the "answer" command, and marking tasks as done only after providing evidence such as test results. The tasks and their associated evidence are stored in the SQLite database, allowing the agent to refer back to them in subsequent sessions.
qtask is compatible with various AI coding agents like Claude Code, Codex, Cursor, Aider, and Gemini CLI, as long as they can run shell commands. A simple setup command is provided for each agent, and any command can be invoked with the "--json" flag to obtain machine-readable output. The tool's underlying SQLite database uses the Write-Ahead Logging (WAL) mode to ensure thread safety, allowing multiple agents to attempt claiming a task without conflicting with each other.
Comprehensive testing is performed using SQLite's full-text search FTS5, with a fallback to a simple LIKE search for builds that do not support FTS5.
The tool is designed to be used with Python 3.11 or newer and git 2.31 or newer, and it can be installed using pipx by running "pipx install git+https://github.com/Abubakr-Alsheikh/qtask". After installation, the user can verify the version with "qtask --version". The developer plans to add features such as converting local tasks into GitHub issues, maintaining synchronization between the two, and creating a read-only HTML board.
The project is open to issues and pull requests, and feedback on compatibility across different agents is welcome.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
