Use Return rather than Linefeed when typing with xdotool

[edef1c]

Most applications don't recognise the Linefeed key (and thus ignore it), so we replace \n with \r, which sends the Return key instead. This makes multi-line input work correctly.

[x5a]

@edef1c we're discussing this internally - thanks for your contribution. We unfortunately cannot accept unsigned commits - see this comment for more details on how to correct.

[x5a]

@edef1c can you share a little bit more about the use case where you're seeing this? in most cases, the model should prefer a computer key <return> action

[edef1c]

@edef1c we're discussing this internally - thanks for your contribution. We unfortunately cannot accept unsigned commits - see this comment for more details on how to correct.

Done.

@edef1c can you share a little bit more about the use case where you're seeing this? in most cases, the model should prefer a computer key <return> action

It doesn't do so consistently. My rather synthetic test was to ask it to open up Jupyter Lab and type some code. It's not consistently reproduced by asking it to open a graphical text editor and type some text with newlines, but it can be made consistent with some coaxing. I'm perfectly happy with the model behaviour here: it seems pointless to spend several tool use cycles to do that.