July 16, 2026
Open source, closed link
Show HN: Sentinel – open-source QA agent that reads your code before it clicks
AI bug hunter wowed people, then the missing repo link stole the spotlight
TLDR: Sentinel says it can study an app like a human tester and already found serious hidden booking errors in a hotel system. But in the comments, the real drama was the broken GitHub link and a brutal joke that framed the launch as polished AI hype with the doors locked.
A shiny new project called Sentinel arrived promising something bigger than a simple button-clicking bot: it says it can read an app’s code first, figure out what the app is actually supposed to do, and then test real customer journeys from start to finish. In the demo, it poked through a hotel booking system, taught itself the major tasks a hotel worker would care about, and uncovered messy problems hiding behind happy-looking screens — including bookings that looked fine on the page but were wrong behind the scenes. For anyone who has ever yelled, “But it worked on my screen!”, this was catnip.
But the comments? Instant chaos. Before the crowd could fully applaud the bug-finding robot detective, multiple people jumped in with the same blunt complaint: the supposedly open-source code link led straight to a 404 error page. That instantly changed the mood from “wow, cool” to “uh, where is it?” The loudest reaction wasn’t even about the software’s clever testing tricks — it was about the old internet classic of launching an “open” project that nobody can actually open.
Then came the drive-by comedy. One commenter zeroed in on the site’s AI-generated blogging slogan and delivered the thread’s sharpest roast, basically accusing the whole thing of being AI slop with a shipping label. So the vibe became a perfect tech-forum cocktail: part genuine interest, part skepticism, and part people mercilessly clowning on the presentation. In other words, Sentinel found bugs in the app, and the community immediately found bugs in the launch.
Key Points
- •Simbastack introduced Sentinel as an MIT-licensed open-source QA agent that reads code to infer product workflows before testing them.
- •The article contrasts Sentinel with UI-focused AI agents, arguing that Sentinel models business logic and validates both frontend and backend behavior.
- •Sentinel was tested on KaribuKit, a hotel property management system built with a Next.js frontend, a separate API service, and a Postgres database.
- •Without a supplied test plan, Sentinel derived nine business flows from the codebase, including reservations, group bookings, rate management, night audit, payments, and guest self-service.
- •The demonstration reported backend and cross-layer bugs, including inconsistent availability state, API/UI disagreement on room availability, and an incomplete check-in state transition despite a 200 response.