im curious how context is managed inside @bot
i can give the same bot completely different tasks, and it somehow keeps enough context between em to work surprisingly well
Is it conversation history, persistent memory, retrieval, summaries, or some other state that gets carried over?
And more importantly: can I tell Grok, “forget everything from this task and start completely fresh”?
any docs i can read to understand how to use this @bot properly?
- im curious how context is managed inside @bot i can give the same bot completely different tasks, and it somehow keeps enough context between em to work surprisingly well Is it conversation history, persistent memory, retrieval, summaries, or some other state that gets carried over? And more importantly: can I tell Grok, “forget everything from this task and start completely fresh”? any docs i can read to understand how to use this @bot properly?
Whatever the implementation, the product should expose a context receipt: task-local history, retrieved memory, summary, and excluded state. “Forget this task” is trustworthy only if the next run can prove which stores were cleared.