AgentProbe – test AI agents by making the systems around them fail

1 pointsposted 13 hours ago
by SamvidK

2 Comments

lucagiftzek

13 hours ago

THat's actually pretty interesting and could make a lot of sense to implement into a real production enviroment. Has anyone used it? what's the results? It catches more than jsut a second chat used as watcher?