Skip to content
Merged
Show file tree
Hide file tree
Changes from all commits
Commits
File filter

Filter by extension

Filter by extension

Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
1 change: 1 addition & 0 deletions github-management/README.md
Original file line number Diff line number Diff line change
Expand Up @@ -19,6 +19,7 @@ Experience Special Interest Group.
- [Repository Creation Guidelines](kubernetes-repositories.md)
- [Setting up the CNCF CLA Check](setting-up-cla-check.md)
- [GitHub Permissions](permissions.md)
- [AI Code Review Tools](ai-code-review-tools.md)

## GitHub Administration Team

Expand Down
99 changes: 99 additions & 0 deletions github-management/ai-code-review-tools.md
Original file line number Diff line number Diff line change
@@ -0,0 +1,99 @@
# AI Code Review Tools

The Kubernetes project may evaluate AI-powered code review tools on a
per-repo opt-in basis. This document describes the process for requesting,
evaluating, and deciding on the use of such tools.

## Scope

This policy covers AI tools that automatically review pull requests, such as
CodeRabbit or GitHub Copilot code review. It does not cover other AI-powered
tooling such as CI/CD, security scanning, or code generation assistants.

Comment on lines +9 to +12
Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

what about open source tools leveraging AI ? we could have something built in the open just for issue triage for example.

Copy link
Copy Markdown
Contributor

@kannon92 kannon92 Apr 6, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

The CNCF provides https://dosu.dev/ for this purpose already. ref: https://contribute.cncf.io/resources/services/hosted-tools/#tools

But imo I'm not sure this policy should be in scope for your question.

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this matters if it's open source or not -- it's more about the role in the software development lifecycle.. specifically code review.

## Requesting a New Tool

A subproject lead or an approver listed in the repository's top-level [OWNERS]
file files an issue on [kubernetes/org]. The issue must:

- Identify the tool and link to its documentation
- Describe the use cases and what the subproject is trying to accomplish
- Explain why existing approved tools do not meet their needs
Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Is there a list of already-approved tools somewhere?

Copy link
Copy Markdown
Member

@janetkuo janetkuo Apr 16, 2026

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

IIRC k8s repo admins can enable Copilot auto code reviews, but as @kannon92 mentioned in kubernetes/org#5930 (comment), it requires PR authors to have Copilot subscription first, or the maintainers need to manually request reviews from Copilot

Copy link
Copy Markdown
Contributor

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

OK. I'm just saying that if we have existing tools, we should link to a list somewhere, so that it'll save a lot of duplicate requests.

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Speaking of this, if a tool is approved in one SIG repo/subproject, can we assume it's approved to use in other repos as well (may still need admin to configure it, but no additional approval is needed)?

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

@janetkuo assuming there is not extra $$$'s involved perhaps?

Copy link
Copy Markdown
Member

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Yes agreed

Copy link
Copy Markdown
Member Author

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Added a blank section for approved tools

- List the specific repositories for the pilot
- Acknowledge the [AI guidance for pull requests]
- List the GitHub permissions and OAuth scopes the tool requests

## Privacy and Security Assessment

Upon receiving a request, the [GitHub Administration Team] conducts a privacy
and security assessment of the tool. The assessment documents:

- What GitHub permissions and OAuth scopes the tool requires
- What data the tool accesses and where it is sent
- What AI models are used to process the code
- Data retention and deletion policies
- Security certifications (SOC2, etc.)
- Whether access can be scoped to specific repositories

The assessment is documented in the [kubernetes/org] issue for transparency.

## Approval

The [GitHub Administration Team] reviews the request and the privacy and
security assessment, and approves or rejects the request. If approved, the
[GitHub Administration Team] enables the tool on the requested repositories
and applies an org-wide default configuration.

## Pilot Structure

- The pilot runs for a 90-day evaluation period starting from the date the
tool is enabled
- The tool is enabled only on the specific requested repositories, not org-wide
- An org-wide default configuration is applied; repositories may customize
within those bounds
- The sponsoring subproject is responsible for collecting feedback from
contributors and reviewers during the pilot

## Evaluation and Decision

At the end of the pilot period, the sponsoring subproject posts an evaluation
summary to the original [kubernetes/org] tracking issue. The summary should
cover:

- Quality of reviews (signal vs noise, with examples)
- Contributor and reviewer feedback
- Any issues encountered
- Recommendation (continue, modify, or remove)

The [GitHub Administration Team], in consultation with the sponsoring
subproject, decides to:

- Continue and expand to additional repositories
- Continue with modifications
- Remove the tool

Expansion to additional repositories requires a new issue on [kubernetes/org].
If the tool has already been approved and assessed, the privacy and security
assessment does not need to be repeated. Requests to enable an already-approved
tool on additional repositories are evaluated on cost impact and repository
fit. Approvals for tools with per-repository costs may require additional
review.

## Approved Tools

The following tools have completed the evaluation process and are approved for
use in the Kubernetes project. Subprojects may request enablement on their
repositories without repeating the privacy and security assessment.

No tools have completed the evaluation process yet. This section will be
updated as tools are approved.

## Removal

If a pilot is unsuccessful or a tool is no longer desired, the
[GitHub Administration Team] will disable the integration. Subproject leads may
request removal at any time by filing an issue on [kubernetes/org].

[GitHub Administration Team]: /github-management/README.md#github-administration-team
[AI guidance for pull requests]: /contributors/guide/pull-requests.md#ai-guidance
[OWNERS]: /contributors/guide/owners.md
[kubernetes/org]: https://github.com/kubernetes/org/issues