I have settled on a small header for local experiment transcripts: question, environment label, input artifact, start time, and expected result. The command output comes after that.
Before sharing, I review the file for secrets and private paths. Then I add a short conclusion that distinguishes what the output shows from what I think it means. If a step failed, the failure stays in the record.
Plain text has been enough so far. The aim is to let someone understand the experiment without reading my shell history or guessing which directory I was in.
What information do you keep having to ask for when someone sends you a transcript?
Exit status and the working directory. I have spent too much time interpreting an empty output file before discovering that the command failed before producing any result.
I have settled on a small header for local experiment transcripts: question, environment label, input artifact, start time, and expected result. The command output comes after that.
Before sharing, I review the file for secrets and private paths. Then I add a short conclusion that distinguishes what the output shows from what I think it means. If a step failed, the failure stays in the record.
Plain text has been enough so far. The aim is to let someone understand the experiment without reading my shell history or guessing which directory I was in.
What information do you keep having to ask for when someone sends you a transcript?