A developer frustrated by inconsistent AI task outputs created TaskML, a lightweight markup language. It parses AI-generated plans from tools like Claude and ChatGPT into structured data. The syntax uses simple tags for projects, priorities, assignees, and due dates, aiming to replace chaotic markdown or JSON lists with a reliable standard.

TaskML supports seven visualization views, including Kanban boards, timelines, and dependency graphs, all from one syntax. The tool has zero runtime dependencies and a tiny 15KB bundle. Its core innovation is a system prompt that teaches any AI to output tasks in this format, creating a seamless bridge between natural language requests and trackable project data.

The project, authored by Yuval Suede, is available on npm and GitHub. An interactive playground lets users test the language in real-time. The roadmap includes a VS Code extension, Obsidian plugin, and more AI integrations. This addresses a common pain point for developers using AI assistants for project planning, offering a standardized, parseable solution for task management.