Merge pull request #3506 from stgraber/llm

Update AI/LLM policy

Author: stgraber

AGENTS.md

@@ -0,0 +1,20 @@
+# Legal
+
+ - All contributions to this repository must be compatible with the Apache 2.0 license.
+ - Specifically (but not limited to), contributions cannot include code licensed under the terms of the GPL, AGPL or LGPL licenses.
+ - Only human beings are allowed to sign the Developer Certificate of Ownership (DCO / Signed-off-by).
+ - Only human beings can ever be credited within commit messages.
+
+# Formatting
+
+ - Code comments should be no longer than one line, unless they are required to cover complex unintuitive logic.
+ - Commit messages should similarly be kept as short and to the point as possible, no need to summarize the whole issue.
+ - We don't use the define and test one line `if` syntax, instead splitting defintion and testing across two lines.
+
+# Testing / validation
+
+ - The commit structure described in `CONTRIBUTING.md` should generally be followed.
+ - All branches are expected to pass `make static-analysis` and `go test -v ./...`.
+ - Excessive unit tests are generally discouraged.
+ - When possible, existing system tests should be extended to cover new features.
+ - A full local system test run isn't required prior to contribution, all tests get run in our CI.

CONTRIBUTING.md

@@ -18,21 +18,83 @@ By default, any contribution to this project is made under the Apache
 The author of a change remains the copyright holder of their code
 (no copyright assignment).
 
-## No Large Language Models (LLMs) or similar AI tools
+## Policy on the use of Large Language Models (LLMs) and AI tooling
+### For issue reporting
 
-All contributions to this project are expected to be done by human
-beings or through standard predictable tooling (e.g. scripts, formatters, ...).
+We do NOT allow direct filing of issues by LLMs.
 
-We expect all contributors to be able to reason about the code that they
-contribute and explain why they're taking a particular approach.
+We REQUIRE a human being to go through our issue reporting form on
+Github and accurately describe their issue and provide all needed
+information.
 
-LLMs and similar predictive tools have the annoying tendency of
-producing large amount of low quality code with subtle issues which end
-up taking the maintainers more time to debug than it would have taken to
-write the code by hand in the first place.
+The more concise and to the point the issue is, the more likely it is to
+be understood, tracked down and resolved quickly.
 
-Any attempt at hiding the use of LLMs or similar tools in Incus contributions
-will result in a revert of the affected changes and a ban from the project.
+Long winded AI written essays can easily look overwhelming and cause our
+maintainers and other contributors to just entirely skip the issue to
+focus their energy on something else.
+
+We also don't benefit from AI generated root cause analysis or proposed
+fixes in those issues. If you yourself understand the code base well
+enough to go through that content and suggested fix, then turn it into a
+pull request and submit it yourself. Otherwise, please limit your report
+to describing the issue at hand and we'll take it from there.
+
+### For contributions
+
+We REQUIRE all contributions to Incus to be submitted by human beings who
+can assert full copyright ownership of their contribution or have been
+allowed by their employer to contribute. This is what the DCO (see below)
+requires of all contributors.
+
+AI tools can sometimes be beneficial, particularly when it comes to
+finding patterns among a large data set (entire code base), performing
+tedious repetitive changes or large refactoring/re-organization.
+
+While we now tolerate the use of such tools, they must abide by our
+instructions (`AGENTS.md`) and their operators cannot override those
+instructions.
+
+We expect everyone contributing to Incus to fully own their
+contribution, be able to reason about it, be able to explain why things
+were done a particular way and act as the full owner of that code. AI
+tools are treated the same as traditional tooling like `sed`, `awk` or
+`coccinelle`.
+
+For the purpose of this project, AI tools CANNOT be treated as author,
+co-author or be credited in any way that would suggest any ownership
+over the contribution.
+
+The contributor should have done all the thinking, planning and
+understanding of the changes needed to resolve an issue or implement a
+new feature prior to using automated tooling to perform the grunt work.
+
+Unguided use of those tools or the inability to prove understanding of
+the code contributed will result in a loss of trust in that contributor
+by project maintainers which can then lead to exclusion from any further
+contribution to the project.
+
+It's also worth pointing out that while those tools are good at
+implementing the more boring/repetitive/grunt work. We've generally
+found that you only really understand the project and its structure by
+having done such work yourself a few times.
+
+### For anyone with write access to the repository
+
+Anyone with write access to this repository must ensure to NEVER run an
+AI agent or similar tool on a system which holds repository credentials
+(SSH key, GPG key, web browser cookies, ...).
+
+Any use of AI tooling should be done inside of a clean VM/container that
+itself cannot directly push to or alter this repository in any way.
+
+The safest approach is to SSH into that environment and then extract the
+changes using `git format-patch`, then review and apply them to your
+actual tree, tweak them as needed, sign them off and then push and open
+the pull request.
+
+Any potential credential compromise or loss of control should be
+immediately reported to `security@linuxcontainers.org`.
 
 ## Pull requests
 

doc/.wordlist.txt

@@ -109,6 +109,7 @@ GIDs
 Github
 Golang
 goroutines
+GPG
 GPUs
 Grafana
 HAProxy