What problem does it solve? Writing feature specifications from scratch and synthesizing scattered PR review comments into actionable changes is slow and error-prone. This Skill automates both steps of a spec-driven development workflow using AI-assisted drafting and feedback analysis. ## Core Features & Use Cases - Draft Proposal Generation: The propose command creates a new directory under docs/changes/ and generates three coherent files (proposal.md, spec-delta.md, tasks.yml) via chained Gemini CLI calls, where each document uses the previous one as context. - PR Feedback Synthesis: The update command fetches review comments from a GitHub PR using the gh CLI, combines them with the existing spec files, and asks Gemini to produce a per-file summary of recommended changes. - Use Case: A product manager wants to propose a real-time notifications feature. They run the propose command, review the generated proposal, spec delta, and task breakdown, open a Spec PR, and later use the update command to turn reviewer comments into a concrete revision plan. ## Quick Start Ask the assistant to create a spec proposal for your feature, for example: propose a change called "Real-time Notifications" using the spec-authoring workflow.