plan

Task decomposition breaks accepted proposals into discrete work items by mapping component boundaries, defining acceptance criteria, and identifying dependencies. You provide the proposal file path, and the system outputs a structured JSON array of tasks with assignees, complexities, and testable done-whens.

Generating a project plan with clear dependencies involves analyzing accepted design artifacts to surface open questions and blocking design tasks. The process maps data shapes and component boundaries to produce concrete deliverables, writing the resulting task array to a JSON file for submission.

Creating testable acceptance criteria from conversation artifacts requires extracting component boundaries and data shapes defined during the design phase. The decomposition process translates these design decisions into discrete tasks with clear done-whens, ensuring engineers and agents understand exact scope without guessing.

Automating task management from proposal files is supported by writing a JSON array of tasks to a temporary directory and submitting them via the task tool. This automation updates proposal frontmatter to a planned state and assigns work items with complexities, related items, and blocking dependencies.

Task decomposition actively surfaces architecture gaps by mapping data shapes and creating blocking design tasks where dependencies exist. The generated output includes open questions and blocked-on fields within the JSON array, ensuring unresolved architectural constraints are tracked before engineering work begins.

The task plan output uses a JSON array format containing structured fields for alias, assignee, complexity, overview, related items, and acceptance criteria. This file is written to a temporary directory path and submitted via the task tool to integrate with existing project workflows.