aibdd-implement

Executing a markdown checklist without skipping steps requires converting each checkbox into a single TodoWrite item and synchronizing state in lockstep. The aibdd-implement approach enforces one-to-one checkbox execution to prevent drift from the source tasks.md file.

TodoWrite state synchronization works by marking leaf checkboxes complete one at a time and immediately updating tasks.md. Parent items are only marked complete after all descendant checkboxes finish, enforcing invariants that completed todo items match checked markdown boxes.

Tracking acceptance criteria progress with markdown checkboxes is supported by turning each checkbox line into a live todo item. This boundary-aware progress tracking judges completion against the checkbox text and immediately marks the corresponding markdown box as done.

Automating ordered task execution from a markdown file is best handled by deterministic checkbox-to-todo conversion. This preserves the structure and order of parent and child dependencies, ensuring no parent task completes before its descendants are fully satisfied.

AI agent implementation tasks drift when execution is not grounded in the written checklist. Enforcing one-to-one checkbox execution with immediate tasks.md updates after each checkbox is judged satisfied prevents the agent from advancing broader work than the current step.

Checklist execution applies to any tasks.md representing acceptance-grounded work broken into ordered, checkbox-level steps. While designed for AIBDD-generated files, it works with any markdown checklist containing parent and child checkbox dependencies.