When the agent hits a step that needs planitator (a localhost web tool), it opens the page in the browser, then blocks waiting for a response. From my phone that's unreachable, so the agent just sits there forever.
In the bot it looks like a normal "agent is running" run — the session is marked busy, the busy message says use /abort to interrupt. But /abort kills the whole run, not the planitator step. I can't send a text reply, can't skip just that step, and there's no hint that the agent is actually waiting on something I can't reach from here.
When the agent hits a step that needs planitator (a localhost web tool), it opens the page in the browser, then blocks waiting for a response. From my phone that's unreachable, so the agent just sits there forever.
In the bot it looks like a normal "agent is running" run — the session is marked busy, the busy message says use /abort to interrupt. But /abort kills the whole run, not the planitator step. I can't send a text reply, can't skip just that step, and there's no hint that the agent is actually waiting on something I can't reach from here.