Fix for AI edit serialisation issues

[dosumis]

@gouttegd reports:

the agent creates some serialisation glitches because it has no idea of the correct order of tags in a OBO file. I suggested that its output should be filtered through a round of robot convert before being submitted as a PR, to erase this kind of serialisation differences.

Could potentially be fixed with instruction to reserialise with ROBOT HERE:

https://github.com/obophenotype/uberon/blob/master/CLAUDE.md#edits

@cmungall - maybe we could replace use of https://github.com/cmungall/obo-scripts/blob/master/obo-checkin.pl with robot merge?

[gouttegd]

When used in ODK-managed ontologies the agent could also be instructed to call the built-in normalize_src target:

make normalize_src

This calls ROBOT under the hood, but dispenses the caller from having to worry about whether the -edit file is in OBO or OFN format. Though maybe it’s not important for now if the agent can currently only work on ontologies whose -edit file is in OBO format.

And +1 for using robot merge (which again would work independently of the -edit file format) rather than a OBO-specific script (with all due respects to Chris’ Perl hacking habits!).

[github-actions[bot]]

This issue has not seen any activity in the past 6 months; it will be closed automatically one year from now if no action is taken.

[dosumis]

@Caroline-99 can you take a look at Claude.md and add if missing?