speckit-tasks

To generate dependency-ordered implementation tasks from user stories, map your plan.md tech structure and spec.md user stories into a strict checklist format with task IDs, file paths, and parallelization markers. Optional artifacts like data-model.md and contracts/ further tailor independent test criteria per story.

The best way to structure tasks.md for AI execution is using a strict checklist format organized by setup, foundational prerequisites, and user-story phases. This dependency-ordered workflow includes completion-order graphs, task IDs, story labels, and file paths to ensure immediate executability.

Yes, you can use optional data-model.md and contracts/ artifacts to tailor independent testable task phases. These design artifacts enhance the core spec.md user stories and plan.md tech structure, providing specific constraints and criteria for each generated task phase.

Before_tasks and after_tasks extension hooks work by running optional pre- and post-generation hook instructions defined in your project's .specify/extensions.yml file. When present, these hooks execute alongside the task generation process within the .specify directory structure.

Generating executable tasks does not require all available Speckit design artifacts. The process mandates plan.md and spec.md, but optionally enhances the output using data-model.md, contracts/, research.md, and quickstart.md to tailor task phases and independent test criteria.

Parallelization markers are needed in a dependency-ordered task checklist to identify independent, testable task phases that can be executed concurrently. They map the completion-order dependency graph from user stories, ensuring foundational prerequisites finish before dependent work begins.