3 out of 4 devs failed to catch dangerous AI-suggested commands

Show HN: 3 out of 4 devs failed to catch dangerous AI-suggested commands

3 out of 4 devs failed to catch dangerous AI-suggested commands is a company in show hn launch.

Profile

Hey HN, I built this. AgentsAegis is a trap-based security training app. Think KnowBe4 for AI agentsThe backstory: I'm a software engineer 14yoe, I use Claude Code daily. Sometimes I approve permission requests and only then read what I just approved Which is ironic as my primary spec is core back-end: security and work with big data. So I built this for myself to not become one of these stories: "Claude ran terraform destroy for production"Concept is simple: Go proxy that sits between AI assistant and the API, intercepts responses, and occasionally swaps in realistic trap commands. If you approve blindly - you get caught. Sounds harsh, but again - I really dont want me or anyone to add 'cau

As a business in show hn launch, 3 out of 4 devs failed to catch dangerous AI-suggested commands is listed here with the public details available about it.

Its current recorded stage is Show HN launch.

Website: agentsaegis.com/assessment