OpenIngress checks whether AI agents can actually get around a website - click the right things, find pages, and finish tasks. Built for teams who care about agent traffic before it becomes a real channel - and for anyone who wants a clear report of what breaks.

hey Product Hunt 👋 we built OpenIngress after auditing ecommerce sites for whether AI agents could actually shop them - not “does it look fine,” but can an agent get to product → cart → checkout without dying on unlabeled controls, drawers, or modal traps. humans usually made it. agents often didn’t. the useful output was always the same: the job, where it broke, what to fix. today we’re open-sourcing that engine. paste a URL, crawl the live site, explore with your LLM key, get coverage + break points. findable ≠ usable. if agents are going to browse for people, you’ll want those failure points before the traffic shows up. self-host it, try your site, tell us what breaks. if you want a custom simulation/build then reach out :)

One thing that would help us a lot is letting us upload a customer journey script and then comparing how different agents perform on that exact path side by side. Right now the tasks feel generic so we have to handcraft each test, which gets tedious when we want to validate changes after every deploy.

OpenIngress

Can AI agents use your site?

Can AI agents use your site?