In 1970 Ted Nelson invented Stretchtext. Basically multi-resolution text. It never caught on because the effort for writing and maintaining such texts is huge. But in the meantime we have LLMs, the cheapest writers ever. Stretchtext could have many purposes, but a very obvious one is within the prompting dialog. Do you trust the summary of your agent after spending hours on a task? Do you really scroll through the whole thing? Wouldn't it be great to zoom in or out of more or less detailed summaries of the dialog? So you can get the bigger picture of what happened, and quickly check the details that matter.
An AI agent can run through a whole lot during hours of execution. Checking the whole journey is not that easy. You have either a summary full of gaps in the end, or a wall of text.
After 56 years, because of LLMS, the time for Stretchtext has finally arrived.
LLM Agent power users (certainly developers)
a harness with built in stretchtext so we can zoom in and out of prompt conversations somehow