| Re: [eclipse.org-architecture-council] AI_POLICY.md document template |
Hi,
I have also read a bit about it and it seems indeed usual to have an
AI_POLICY.md but as you said it seems not being a standard, so mostly
for humans (that tend to ignore such instructions anyways) so for it
being effective I think the key is that it is discoverable (and will be
used) by the *AI agent* itself.
I then asked copilot
> I recently read about AI_POLICY.md and want to know in what cases you
will use/follow the content of such file.
>> AI_POLICY.md isn't one of my recognized instruction files.
>> Copilot CLI only automatically reads instructions from:
>>
>> • CLAUDE.md, GEMINI.md, AGENTS.md (git root & cwd)
>> • .github/instructions/**/*.instructions.md
>> • .github/copilot-instructions.md
>> • $HOME/.copilot/copilot-instructions.md and
>> $HOME/.copilot/instructions/**/*.instructions.md
>> • Dirs listed in COPILOT_CUSTOM_INSTRUCTIONS_DIRS
>> So an AI_POLICY.md file wouldn't be picked up automatically unless
>> it's referenced/linked from one of those files, or you explicitly
>> paste/mention its contents in a prompt (e.g., via @AI_POLICY.md ).
So from the reading of the template it also seems to be mostly directed
to the human reader. In general it feels to be to exhaustive (much text
to read) and one wonders if yet another file helps or if having a really
short summary referencing the existing doc is not sufficient enough?
I have lately tried another approach, that is directed at having it
readable for the agent itself see [1] in the agents and contribution
docs + PR templates. Maybe AI_POLICY.md can be an addition here, but as
long as it lacks Github / Tool integration its likely ignored exactly in
these cases where it matters.
best
Christoph
[1] https://github.com/eclipse-tycho/tycho/pull/6227
Am 17.08.26 um 23:07 schrieb Wayne Beaton via
eclipse.org-architecture-council:
> I don't know specifically where Maria Teresa got her inspiration for the
> template that she's generated.
>
> We started this work based on an observation that AI_POLICY.md documents
> in various forms are popping up in all kinds of GitHub repositories.
> AFAICT it's not a standard yet, nor have I seen any specific discussion
> regarding making it a standard. Things continue to evolve.
>
> If you (or anybody else in the council) has observed something
> different, then we should explore that.
>
> Here're a couple of examples:
>
> https://github.com/eclipse-collections/eclipse-collections/blob/master/
> AI_POLICY.md <https://github.com/eclipse-collections/eclipse-
> collections/blob/master/AI_POLICY.md>
> https://github.com/apache/fory/blob/main/AI_POLICY.md <https://
> github.com/apache/fory/blob/main/AI_POLICY.md>
>
> HTH,
>
> Wayne
>
> On Mon, 17 Aug 2026 at 16:08, Jay Jay Billings via eclipse.org-
> architecture-council <eclipse.org-architecture-council@xxxxxxxxxxx
> <mailto:eclipse.org-architecture-council@xxxxxxxxxxx>> wrote:
>
> Wayne,
>
> Thanks very much for sharing. I look forward to digging deeper into
> this.
>
> Quick question: Can you point to an example of a project that has an
> AI_POLICY.md document? I've never seen that kind of document.
> Instead, I usually see the AI policy in README.md.
>
> Jay
>
> On Mon, Aug 17, 2026 at 3:18 PM Wayne Beaton via eclipse.org-
> architecture-council <eclipse.org-architecture-council@xxxxxxxxxxx
> <mailto:eclipse.org-architecture-council@xxxxxxxxxxx>> wrote:
>
> Greetings Eclipse Technical Advisory Council!
>
> We've observed a trend of creating an AI_POLICY document in
> repository roots to provide contributors with guidelines
> regarding the project's values with regard to the use of AI in
> the contribution process. I've mentioned this document in a few
> of our committer office hours sessions and in the Use of
> Artificial Intelligence by Eclipse Committers and Contributors
> <https://www.eclipse.org/projects/training/#ai> committer
> training video.
>
> The EMO team has been working on a template for a project-
> specific AI_POLICY.md file. We're tracking that work in this
> issue <https://gitlab.eclipse.org/eclipsefdn/emo-team/emo-
> services/-/work_items/32>. Your feedback is requested.
>
> Note that this is intended as a template for a project-specific
> policy; this is specifically not intended as the policy for all
> Eclipse projects. Nonetheless, this will likely be regarded as a
> recommendation by some, so we should be sensitive to that.
>
> We considered the opportunity for multiple templates but decided
> to start with what the team feels is the most common configuration.
>
> Thanks in advance.
>
> Wayne
>
> --
>
> Wayne Beaton (he/him)
>
> Head of Open Source Projects | Eclipse Foundation
>
> _______________________________________________
> eclipse.org-architecture-council mailing list
> eclipse.org-architecture-council@xxxxxxxxxxx
> <mailto:eclipse.org-architecture-council@xxxxxxxxxxx>
> To unsubscribe from this list, visit https://www.eclipse.org/
> mailman/listinfo/eclipse.org-architecture-council <https://
> www.eclipse.org/mailman/listinfo/eclipse.org-architecture-council>
>
>
>
> --
> Jay Jay Billings, Ph.D.
> _______________________________________________
> eclipse.org-architecture-council mailing list
> eclipse.org-architecture-council@xxxxxxxxxxx <mailto:eclipse.org-
> architecture-council@xxxxxxxxxxx>
> To unsubscribe from this list, visit https://www.eclipse.org/
> mailman/listinfo/eclipse.org-architecture-council <https://
> www.eclipse.org/mailman/listinfo/eclipse.org-architecture-council>
>
>
>
> --
>
> Wayne Beaton (he/him)
>
> Head of Open Source Projects | Eclipse Foundation
>
>
> _______________________________________________
> eclipse.org-architecture-council mailing list
> eclipse.org-architecture-council@xxxxxxxxxxx
> To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/eclipse.org-architecture-council
_______________________________________________
eclipse.org-architecture-council mailing list
eclipse.org-architecture-council@xxxxxxxxxxx
To unsubscribe from this list, visit https://www.eclipse.org/mailman/listinfo/eclipse.org-architecture-council