outline-expand

You can automate research paper drafting by dispatching a dedicated writer agent per section in a single parallel batch. This approach generates first drafts of all unapproved manuscript sections concurrently from a pre-defined outline, reducing sequential writing time to minutes.

To maintain terminology consistency across drafted sections, an identical shared nomenclature payload is passed to all parallel writers. An automated drift lint then scans the outputs for inconsistent jargon and generates a report to surface discrepancies for user reconciliation.

Before generating first drafts, you need a completed paper outline and an initialized paper.json state file. These inputs provide the structural framework and project state required to dispatch the parallel agents for manuscript expansion.

Yes, you can draft all unapproved manuscript sections in a single parallelized run. The map-reduce mechanism dispatches a dedicated paper-writer agent for each section concurrently, eliminating the need for manual sequential writing of individual sections.

If jargon drift occurs in your drafted research paper, an automated post-draft scan generates a non-blocking terminology drift report. This report surfaces inconsistencies across sections so you can easily reconcile project-specific jargon without blocking the drafting process.

The best way to refine individual sections after parallel research paper drafting is to use an iterate-revision skill. Once the outline-expand skill generates the concurrent first drafts, you can transition to refining each section individually to finalize the manuscript.