What WhatProblem does when an issue is opened, how far it goes, and everything it leaves alone.
Open an issue. The bot reads it and replies within a minute or two with the questions it needs answered before anyone could pick the issue up, along with a rough clarity score.
You do not have to answer. If the issue was already clear, the bot says so and leaves it alone rather than inventing questions to justify the comment.
It will not go past five rounds on one issue. After the fifth, the next comment gets the closing summary instead of another question.
On an issue that arrived clear, step five is a short closing note rather than a summary, because restating a well-written issue back to its author is not worth a second comment.
Nothing happens at the one week mark itself. There is no timer running against your issue and nothing gets labelled.
The next comment on the issue is what checks. If the thread has been quiet for more than a week by then, the bot treats the conversation as finished. That comment is not analysed, but it does get one reply saying what happened and how to pick the issue up again, so you are never left wondering whether the bot broke.
A week is deliberately longer than it takes to answer a question properly. Fetching a log file, reproducing something, or coming back on Monday should never be the thing that ends the conversation. When you do pick it up with !whatproblem analyze, the bot re-reads the whole issue including anything posted in the gap, so nothing you wrote is wasted.
Two commands cover everything: one to call the bot onto an issue, one to make it leave. Both are on the commands page, along with the older forms that still work and what stopping does and does not apply to.
It only ever posts comments. It does not close your issue, label it, edit anything you wrote, or open pull requests.
Each response carries a rough percentage. It is the bot's own estimate of how ready the issue is for someone to pick up, not a grade on your writing, and it is not used for anything except deciding whether to ask another question.
A low score on a first pass usually means context that is obvious to you has not made it into the text. That is the normal case rather than a failure: the things a reader needs are exactly the things the person closest to a problem stops noticing.
One issue costs one against your monthly allowance, however many rounds of conversation it takes. Plans and limits has the detail, including what is never charged.
Most of the time it is one of the exclusions above. If none of those fit, the troubleshooting list works through the rest, and it sits next to the tracker so you can file the problem from the same page if none of it helps.