Inspiration
Data breaches and leaks from e-commerce sites have huge and far-reaching impacts (such as Target in 2013/2014 and V-Tex in 2025). These are often a result of small oversights in coding and security logic. Why not make a tool to automatically stress-test e-commerce sites for vulnerabilities and suggest patches?
What it does
Paste your e-commerce site URL into our website. Our agent then stress-tests the site to find any security vulnerabilities and bugs. These are listed on our site's dashboard, where you can review the issues and let our AI agent attempt to automatically fix them.
How we built it
- We used vercel / next.js / typescript for the frontend
- AWS agent core
- AWS strands for the agent framework with playwright as Browser Driver
- juice-shop for the sample
- python for backend
- An endpoint on the next.js website launches the agent on AWS agent-core with the given website
Challenges we ran into
Nothing works! But actually...teammates eat too much pizza. But really...AWS didn't give a sandbox (just API key), had to create own AWS account. When creating account, page says that the verification code can take up to 5 min, but countdown only goes to 60s. Wanted to use AWS agent core browser tool, but documentation is incomplete (wasn't clear that the CLI/agent core is from the agent core starter kit tool). Have to install starter toolkit get actually get agent core CLI.
Accomplishments that we're proud of
Being able to run Playwright with Strands
What we learned
How to run Playwright with Strands
What's next for ShopSentry
Built With
- amazon-web-services
- claude-agent-sdk
- github
- next.js
- python
- react.js
- supabase
- vercel
Log in or sign up for Devpost to join the conversation.