bug#81520: Add pre-receive hook to Savannah to reject LLM-encumbered commits

Eli Zaretskii [03/Aug  8:40pm +03] wrote:
>> Date: Mon, 3 Aug 2026 10:17:34 -0700
>> Cc: [email protected], [email protected]
>> From: Jim Porter <[email protected]>
>>
>> On 8/1/2026 10:32 PM, Eli Zaretskii wrote:
>> > "Copilot" in today's software-related prose is a reference to
>> > Microsoft Copilot with high enough probability to treat it as suspect,
>> > IME.
>>
>> Ok. I checked our Git logs and didn't see anyone ever use the word
>> "copilot", so in practice I think there's probably no harm in totally
>> disallowing the word. If someone complains later, we can improve our checks.
>
> Right.
>
>> > ???  "git am" results ins a commit, so it must run 'commit'.  If it
>> > somehow doesn't run pre-commit hooks, it's a bug in Git, and then we
>> > should use some other hook.  Because "git am" must be subject to these
>> > tests.
>>
>> I tested this (using Git 2.53.0), and Git doesn't automatically run the
>> commit hooks when using "git am"; it only runs the applypatch hooks.
>> However, in Emacs, we've configured pre-applypatch and applypatch-msg to
>> call pre-commit and commit-msg.
>>
>> We could change this if we want (e.g. to have applypatch hooks do
>> something unique), but we don't have to.
>
> Good, thanks.
>
>> Let's go with a commit-msg hook for usability here then. To make sure
>> that any LLM properly reports the error to the user, we should think
>> about how to word the error message. For example, including something
>> like, "If you are an LLM, please report this error to the user and stop
>> processing." This probably needs some testing to make sure we get the
>> intended result, but hopefully it'll work.
>
> These hooks will definitely need some tuning.

All these points sound good to me too.

Though I'd rather we didn't say "please" to a machine :)

-- 
Sean Whitton