A single AI prompt turns agents into a product design review team that screenshots and fixes UI issues
vista8 · x · 2026-11-09
Borrowing the design QA walkthrough practice from big tech, the author wrote an "AI walkthrough" prompt that works remarkably well.
The prompt casts the AI as a hybrid of product manager, UX researcher, and top UI designer, proactively reviewing every page and interaction without waiting for instructions: it audits information architecture, user flows, operation efficiency, feedback mechanisms, visual hierarchy, and component consistency; triages issues by severity, sending high-priority ones to sub-agents for fixes, and turning the rest into actionable plans.
Screenshots show the agent capturing screen after screen (onboarding, editor pages, export flow), completing a second round with 10+ triaged issues, and waiting for the release build to finish before dispatching sub-agents against the project's UX goals doc.
More from coding & agent
- PM open-sources MIT-licensed agent skills collection for shipping products without coding — PawelHuryn · 2026-11-09
- Agentic AI is orchestration of model calls; datacenter limits may push edge models — DavidLinthicum · 2026-11-09
- Verification is the next bottleneck for AI coding agents, says engineer — StewartalsopIII · 2026-11-09
- athenahealth MCP Server Brings AI Clinical Decision Support With HIPAA-Compliant Workflows — modelcontextprotocol · 2026-11-09
- Google Docs MCP Pack Lets AI Agents Read, Create and Edit Docs via OAuth — modelcontextprotocol · 2026-11-09
- Economist proposes AI-assisted peer review: model reads appendices, expert judges critiques — paulnovosad · 2026-11-09