Skip to content

Verify First Review

Use this page to confirm that the first DiffPal run produced a complete review, even when it found no issues.

Pull Request Signals

The PR or MR must show:

  • a DiffPal Review Summary review, discussion, or thread;
  • inline comments, discussions, or PR threads when actionable findings exist;
  • a check, commit status, or PR status for the DiffPal review job.

Required Artifacts

The CI workspace must contain:

PathPurpose
.artifacts/diffpal/findings.jsonCanonical structured findings bundle.
.artifacts/diffpal/summary.mdHuman-readable review summary.

Depending on the host and feedback mode, the run may also write:

PathPurpose
.artifacts/diffpal/diffpal.sarifSARIF report when enabled by the platform output.
.artifacts/diffpal/codequality.jsonGitLab Code Quality report.

No Findings Vs Broken Review

No findings is healthy when:

  • the review summary appears;
  • findings.json exists and is valid JSON;
  • the CI job succeeds unless a configured gate blocks it;
  • logs show DiffPal resolved the review scope and completed publishing.

A broken review usually has one of these symptoms:

  • no summary appears in the PR/MR;
  • findings.json is missing;
  • provider authentication failed;
  • the workflow used a shallow checkout and could not compare base and head;
  • the platform token could not publish feedback.

Gate Behavior

When gating is enabled, DiffPal fails the CI job if at least one finding meets diffpal.gate.block_on. Setup, authentication, review scope resolution, and publishing failures also fail the job because the review result is incomplete.

When gating is disabled, blocking findings can still be published as review feedback, but they do not fail the CI job. Tooling failures still fail because DiffPal did not produce a complete review.

Troubleshooting

After the first run is healthy, continue with Next Steps.

Released as open source software.