of course it does, that's the point of it.

The notable discovery here isn't that the AI agent has access to its own transcript, it's that 20% extra tokens are used to generate the status line. Leave it to LLM writing to focus on something inconsequential though.