A cursory Google search lead me to this[1], which seems to be what you're looking for, something similar probably exists for ledger/hledger.
Personally having dealt with crypto in beancount a few times I used a separate file which got updated periodically with a script, which worked well enough.
If you're interested I suggest reading the beancount cookbook[2] as it contain many examples.
While that is a very funny sentence, the half a cent I'm withholding from the artist by blocking the ad is hardly a robbery.
Just buy a CD once in a while if the guilt is too much to handle.
I don't think it's a good educated guess though, as ChatGPT itself points out the pronoun is supposed to refer to the closest noun, and while it's reasonable to assume that it ignored technically proper grammar in favour of the more likely option, this still doesn't explain why it insisted the sentence wouldn't make sense if we assumed the lawyer was a woman.
Also while it would most likely not directly say that women can't be lawyers, its answer to the correction only makes sense if we are already under the shared assumption that they can't, which I'd argue is no better.
Finally while this type of idiosyncrasies would be fine and arguably desirable if ChatGPT was only a "big statistical model", they have no place in a product with actual practical applications outside of research.
Personally having dealt with crypto in beancount a few times I used a separate file which got updated periodically with a script, which worked well enough.
If you're interested I suggest reading the beancount cookbook[2] as it contain many examples.
[1] https://github.com/davidastephens/beancount-plugins#automati...
[2] https://beancount.github.io/docs/command_line_accounting_coo...