spec-from-conversation

Provide the conversation transcript along with project context and the desired output format. The skill identifies key topics, extracts requirements and decisions, and organizes them into structured sections like Overview, Requirements, Design, and Action Items.

Requirements extraction works by reading the full conversation, identifying stated goals, features, and constraints, then listing them as functional and non-functional requirements. Technical details like libraries, schemas, and endpoints mentioned in the discussion are preserved.

Yes, open questions and ambiguities are explicitly marked rather than resolved silently. The output clearly distinguishes decided items from unresolved topics so stakeholders know what still needs discussion.

You need the conversation transcript, relevant project or domain context, the desired output format such as spec or ticket, and optionally key topics or stakeholders to focus on. More context produces more accurate and complete specifications.

The output only reflects what was actually discussed and will not invent missing requirements. Gaps in the conversation appear as open questions, so critical details never mentioned in chat must be added separately.