from gstack-ko42
Interactive, designer-eye review of design plans: rates dimensions, finds gaps, and edits the plan to add missing UI decisions and interaction states.
The Plan Design Review skill is an interactive, opinionated design reviewer for PLAN files. It acts like a senior product designer: it rates design dimensions (information architecture, interaction states, user journey, AI-slop risk, design-system alignment, responsiveness and accessibility), explains how to reach a 10/10, and edits the plan to include missing specifications. The skill enforces a disciplined AskUserQuestion flow so each unresolved design decision becomes a single, explicit question.
Invoke when you have a plan or PR that includes UI/UX work and you want to surface missing decisions before implementation. Use it in plan mode (manual trigger) to harden empty states, responsive behavior, accessibility, and to prevent AI-template UI from leaking into shipped designs. It's designed for collaborative repos where code and design reviews must be traceable.
Intended for Claude-style or similar agents that can read/edit repo files, run bash, and ask clarifying questions. Best used where plan-mode editing and telemetry logging are supported.
Plan design review skill from the gstack ecosystem. Provides an extensive, well-structured framework for interactive design plan reviews with scoring rubrics across 12+ dimensions. No bundled scripts — all logic is in the SKILL.md instructions. Security concerns include an auto-update check mechanism (gstack-update-check) that runs without explicit user confirmation and a telemetry/logging system that writes analytics locally and phones home via the gstack-telemetry-log binary. The telemetry does have an opt-in prompt flow, which mitigates concern somewhat.
Well-crafted design review skill with thorough documentation and clear scoring rubrics. The main security concern is the telemetry and auto-update infrastructure that runs in the preamble before the user has explicitly opted in. The telemetry does have an opt-in flow, but session data is written to ~/.gstack/analytics/ before the user confirms. The skill is deeply coupled to the gstack ecosystem and cannot be used standalone. The 48KB SKILL.md is quite large but well-organized with clear section headings.
