design-review

Audits live sites for visual inconsistencies and fixes them with atomic commits and before/after screenshots.

1|Updated Jul 23, 2026
One-click install
npx skills add https://github.com/sanjanb/my-agent-harness --skill design-review-sanjanb
Or copy as Structured Prompt for Agent
Please help me install this Agent Skill.
Skill: design-review
Source: https://github.com/sanjanb/my-agent-harness/tree/main/skills/gstack-design-review
Command: npx skills add https://github.com/sanjanb/my-agent-harness --skill design-review-sanjanb

SYSTEM DOCUMENTATION & REQUIREMENTS

What problem does it solve? Visual QA of a live site is tedious and subjective: spacing drift, hierarchy problems, AI slop patterns, and slow interactions slip through code review. This Skill gives the agent a designer's eye, finding visual issues and fixing them directly in source code with verifiable before/after evidence. ## Core Features & Use Cases - Visual inconsistency detection: Finds spacing issues, hierarchy problems, AI slop patterns, and slow interactions on a live site. - Iterative fix loop: Fixes each issue in source code, commits atomically, and re-verifies with before/after screenshots. - Proactive triggering: Activates when asked to "audit the design", run "visual QA", or polish the look of a live site; pairs with /plan-design-review for pre-implementation review. - Use Case: After shipping a new landing page, ask the agent to audit the design; it screenshots the page, flags misaligned spacing and weak visual hierarchy, fixes each in code, commits per fix, and shows before/after proof. ## Quick Start Ask the agent to audit the design of the running site and fix any visual inconsistencies it finds.

Frequently Asked Questions about design-review

High-intent search queries and answers about installing and using this skill.

FAQPage Schema
How do I run a visual design audit on my live site with an AI agent?

Invoke the design-review skill against the running site. It screenshots pages, identifies spacing, hierarchy, and interaction issues, then fixes each in source code with an atomic commit and re-verifies with before/after screenshots.

What is the difference between design-review and plan-design-review?

design-review audits and fixes a live, implemented site iteratively. plan-design-review runs in plan mode before implementation, reviewing the design approach before any code is written.

What kinds of visual issues does design review detect?

It detects visual inconsistency, spacing problems, hierarchy issues, AI slop patterns, and slow interactions. Each finding is fixed in source code and verified with before/after screenshots rather than just reported.

Does design review work in headless or spawned agent sessions?

Yes. In spawned sessions it auto-chooses recommended options without interactive prompts and ends with a completion report. In headless sessions it blocks when a user decision is required, since no human can answer.

When should I not use automated design review?

Avoid it for pre-implementation design decisions, where plan-design-review fits better, and when the site cannot be run locally for screenshots, since the fix loop depends on visual re-verification.