ShipCheck is my hackathon entry: a pre-publish quality checker for Framer sites, built with Framer 3.0 agents.
Framer 3.0 agents mean anyone can build a site in minutes. But that creates a new problem: anyone can also publish something with weak hierarchy, low contrast, or no clear CTA — and never know it. I wanted to build the thing that catches that before it ships.
Paste a URL and ShipCheck runs an analysis across four areas: visual design, UX, conversion, and accessibility. It returns a quality score, a list of specific issues ranked by severity, and a ready-to-paste prompt for Framer Agent to fix them.
The analysis is real, not simulated — a Vercel backend calls Claude to generate the score and issues for whatever site you give it.
To prove the concept, I ran ShipCheck on its own landing page, copied the prompt it generated, and handed it back to Framer Agent. The page you can compare against itself is the result — same tool, used on itself, closing the loop from audit to fix.
Framer Agent built the page shell and components fast, which let me spend my time on the part that actually matters: the logic connecting the input, the analysis, and the result. Agents are very good at structure and styling; wiring real API calls underneath still benefits from a steady hand.
Try it: paste any Framer site URL, hit Analyze, and copy the fix straight into your own Agent chat.
Built with Framer 3.0 AI agents + Claude
ShipCheck is my hackathon entry: a pre-publish quality checker for Framer sites, built with Framer 3.0 agents.
Framer 3.0 agents mean anyone can build a site in minutes. But that creates a new problem: anyone can also publish something with weak hierarchy, low contrast, or no clear CTA — and never know it. I wanted to build the thing that catches that before it ships.
Paste a URL and ShipCheck runs an analysis across four areas: visual design, UX, conversion, and accessibility. It returns a quality score, a list of specific issues ranked by severity, and a ready-to-paste prompt for Framer Agent to fix them.
The analysis is real, not simulated — a Vercel backend calls Claude to generate the score and issues for whatever site you give it.
To prove the concept, I ran ShipCheck on its own landing page, copied the prompt it generated, and handed it back to Framer Agent. The page you can compare against itself is the result — same tool, used on itself, closing the loop from audit to fix.
Framer Agent built the page shell and components fast, which let me spend my time on the part that actually matters: the logic connecting the input, the analysis, and the result. Agents are very good at structure and styling; wiring real API calls underneath still benefits from a steady hand.
Try it: paste any Framer site URL, hit Analyze, and copy the fix straight into your own Agent chat.
Built with Framer 3.0 AI agents + Claude
