plan-first

To generate an executable task plan from a design document, the system reads your design file and applies deterministic task decomposition to break it down into granular, testable tasks with clear acceptance criteria. The final topologically sorted plan is then written directly into a plan markdown file.

Dependency-based task ordering in project planning organizes tasks topologically based on their prerequisites and prioritizes high-risk items. This scheduling mechanism ensures that foundational tasks are completed before dependent ones, enabling deterministic execution and preventing workflow bottlenecks.

To break down a project design into a checklist with acceptance criteria, automated task decomposition parses the design content and maps each item into a structured format. This format combines the action, goal, and acceptance criteria while recording involved files and prerequisites.

Yes, you need a specific design file format to use automated task decomposition. The system requires the project design to be available as a markdown file stored at the designated dot ai state design path, which it reads to extract and schedule the tasks.

Yes, you can validate a generated project plan before execution. The system implements plan validation as part of its core processing logic to ensure the dependency-based ordering and task decomposition are structurally sound before writing the final output.

The best way to schedule high-risk project tasks using design documents is through dependency-aware scheduling. This approach topologically orders tasks based on their dependencies and explicitly prioritizes high-risk items early in the generated plan.