Summary
When show-git-changes.py reviews updates that only change pinned plugin commits, the current AI-facing diff can degenerate into little more than old-hash/new-hash churn. That is not enough context to decide whether the update is routine or risky.
Problem
For plugin pin updates, the visible change in the consuming repository may be only a commit hash replacement. Reviewing that in isolation is low-value.
A better review surface would include upstream context for the pinned range, such as:
- commit subjects between the old and new pinned revisions
- a short log or summary of the range
- where practical, a focused diff for the plugin content that changed
Goal
Make AI-assisted review of pinned plugin updates meaningfully describe what changed upstream, rather than only repeating the consuming repository's hash change.
Done looks like
- Plugin pin bumps are detected as a special review case, or otherwise get richer upstream context.
- The review input includes at least commit-message context for the old..new range.
- Where practical, the review can also surface a diff or shortlog summary for the plugin changes.
- The resulting AI review is substantially more informative than "hash A changed to hash B".
Non-goals
- Fetching arbitrary large histories without bounds.
- Replacing the normal review path for non-plugin update cases.
Notes
This is specifically about improving the review surface when the downstream diff is too thin to be meaningful on its own.
Summary
When
show-git-changes.pyreviews updates that only change pinned plugin commits, the current AI-facing diff can degenerate into little more than old-hash/new-hash churn. That is not enough context to decide whether the update is routine or risky.Problem
For plugin pin updates, the visible change in the consuming repository may be only a commit hash replacement. Reviewing that in isolation is low-value.
A better review surface would include upstream context for the pinned range, such as:
Goal
Make AI-assisted review of pinned plugin updates meaningfully describe what changed upstream, rather than only repeating the consuming repository's hash change.
Done looks like
Non-goals
Notes
This is specifically about improving the review surface when the downstream diff is too thin to be meaningful on its own.