doc-coauthoring

The doc-coauthoring skill provides a systematic framework for users and Claude to co-author complex structured documents. It is specifically designed for technical writers, product managers, engineers, and leadership teams who need to draft high-stakes content such as PRDs (Product Requirement Documents), technical specifications, design documents, decision documents (RFCs), and formal project proposals. By breaking the writing process into three distinct, manageable stages, this skill ensures that documentation is not only comprehensive but also optimized for reader clarity and comprehension.

• Multi-stage guided collaboration: Moves from intensive Context Gathering to structured Refinement & Structure, concluding with Reader Testing.

• Context-aware knowledge synthesis: Facilitates brain-dumping and meta-context collection, allowing the agent to ask targeted, gap-filling questions to understand project background, audience impact, and organizational constraints.

• Iterative section drafting: Employs a collaborative brainstorming and editing methodology where sections are built step-by-step with continuous user feedback loops.

• Predictive readability verification: Uses a 'fresh Claude' approach for reader testing, simulating an outside reader to identify blind spots, jargon traps, or missing logical flows before the document is finalized.

• Tool-integrated workflows: Leverages available MCP integrations (such as Slack, Google Drive, or SharePoint) to pull relevant project context directly from team resources.

• Trigger this skill by mentioning intents like writing a spec, drafting a proposal, creating a decision doc, or performing a substantial writing task.

• For best results during the Context Gathering phase, provide access to existing team channels, linked documents, or shorthand notes; the agent is designed to process unstructured input efficiently.

• When building sections, prioritize unknowns first (such as the core proposal or technical architecture) before moving on to supporting sections or summaries.

• During the refinement stage, the agent will offer 5-20 brainstorming options to help you choose the most effective structure, tone, and content density for your target audience.

• If images or diagrams are involved, the agent can assist in generating descriptive alt-text to ensure that your technical documentation remains accessible and understandable when read by AI or other team members.