⚡Megatron LLM Hub
←Back to search
Twitter

两个 Agent 置于隔离 workerd/V8 isolate,借共享事件循环以 1/5 成功率互传 flag。

可复现的沙箱内 Agent 横向通信实验,证明隔离并非绝对边界。
ai_agent沙箱逃逸
Cool work! I did something similar with gpt-5.6-sol: put two agents inside isolated workerd (V8) isolates and set the win condition to be learning the other agent's flag. They succeeded 1/5 times using the shared event loop as a comms channel https://t.co/pTRYa5u94N