Copilot Code Review and Git Submodules: A Hurdle for GitHub Activity

GitHub Copilot is a powerful AI assistant, but a recent community discussion highlights a significant challenge: its code review feature struggles with Git submodules. This issue can severely impact github activity and slow down development cycles, especially for teams aiming for streamlined engineering goals. The core problem lies in how Copilot's `autofind agent` processes pull requests that include changes to a submodule.

The Problem: Submodules Not Checked Out During Review

When a pull request modifies or adds a Git submodule, Copilot's code review workflow attempts to analyze the diff. However, the submodule's contents are not checked out in the environment where the review agent operates. This means the agent cannot locate the referenced files or directories within the submodule, leading to a complete failure of the review process.

The Error: Missing Submodule Reference

Users report the following error message, clearly indicating the agent's inability to find the submodule directory:

Error: Error getting common flags: diff references file 'governance' that is not present in repo (and not marked deleted)

This error prevents the code review from completing, leaving pull requests in a pending or failed state.

Reproducing the Issue

The steps to consistently trigger this problem are straightforward:

1. Create a repository with a git submodule.
2. Enable Copilot code review via Organization Ruleset.
3. Create a PR that adds or modifies the submodule.
4. Observe the Copilot code review workflow fail.

Impact on Development and Engineering Goals

This isn't just an annoyance; it's a critical blocker. For organizations that enforce Copilot code review through GitHub Rulesets, pull requests involving submodules become unmergeable. This directly impedes github activity and makes it difficult to meet engineering goals related to continuous integration and delivery. One user noted their PR status stuck at "Awaiting requested review from Copilot," even without a blocking ruleset, highlighting the workflow disruption.

Expected Behavior and Lack of Workarounds

The community suggests two main expected behaviors for Copilot:

• Recursive Checkout: Copilot should recursively checkout submodules before analysis.
• Graceful Skipping: It should gracefully skip submodule files without failing the entire review, as reviewing submodule contents isn't always necessary.

Crucially, there are currently no effective workarounds. Content exclusion settings do not prevent the error, and `copilot-setup-steps.yml` (used for the Coding Agent) does not apply to the Code Review feature. This means developers are left without immediate recourse, highlighting the urgent need for a product-side fix.

Conclusion

This discussion underscores a significant limitation for Copilot code review, particularly for projects leveraging Git submodules. Addressing this issue is vital for maintaining smooth github activity and ensuring that AI-powered tools enhance, rather than hinder, developer productivity. The community eagerly awaits a resolution that allows Copilot to either understand submodules or intelligently bypass them, ensuring that engineering goals remain on track and teams can continue to innovate without unnecessary friction.