doc-coauthoring

Start by stating you want to write a design doc, and the workflow begins with context gathering where you dump background information and answer clarifying questions. The document is then built section by section through brainstorming and iterative drafting, followed by reader testing.

If you don't know what sections you need, the workflow suggests 3-5 sections appropriate for your doc type, such as problem statement, proposal, and alternatives. You can adjust the structure before drafting begins, starting with the section that has the most unknowns.

Yes, if integrations like Slack, Google Drive, or other MCP servers are available, the workflow can pull context directly from shared documents and team channels. Without integrations, you can paste content manually or enable connectors in your settings.

Reader testing validates your document by asking a fresh reviewer with no prior context to answer questions real readers would ask. It exposes ambiguity, assumed knowledge, and contradictions that authors typically miss because they already understand the material.

Skip it for short or informal writing where three stages of iteration add unnecessary overhead. You can decline the workflow at the start and work freeform, or skip individual stages if you prefer faster drafting.