deepen-plan

To fill in TBD and TODO placeholders in a feature plan, you can use a research workflow that scans plan.md, iterates through up to three research passes, and replaces vague sections with actionable implementation guidance before stakeholders review it.

The best way to research unclear feature plans is to apply targeted research passes that find unclear parts and expand them with specific supporting detail, hardening the plan to reduce ambiguity and risk before the next review step.

You need to harden a feature plan before red team review when key sections are still uncertain, vague, or under-researched, ensuring the review step starts from a more concrete foundation with open questions about implementation, rollout, or edge cases resolved.

To replace needs clarification sections in plan.md, scan the file for vague language and iterate through research passes, committing after each pass to report cumulative results that replace placeholder language with concrete implementation guidance.

Yes, this feature planning workflow is designed for complex features with edge case gaps, expanding uncertain plans into concrete implementation guidance through up to three targeted research passes to fill gaps before stakeholders review.

A limitation of automated research passes is that the workflow is capped at three iterations per plan, requiring sufficient context in plan.md to expand vague sections, and it is best used post-planning rather than for initial feature drafting.