posting-review-summary

Posting Review Summary

Context Detection

Check contexts in this order — use the first match:

FORBIDDEN: Do not use gh pr comment to create summary comments.

PR Metadata Assessment

If PR title, description, or test plan is genuinely deficient, add as a finding in the Code Review Details collapsible section.

Rules

• DO NOT comment on minor improvements
• DO NOT comment on adequate-but-imperfect metadata
• NEVER add as an inline comment
• DO NOT exceed 3 lines of feedback on the PR Metadata Assessment

Examples

Genuinely deficient means:

• Title is literally "fix bug", "update", "changes", or single word
• Description is empty or just "See Jira"
• UI changes with zero screenshots
• No test plan AND changes are testable

Adequate (DO NOT flag):

• Title describes the change even if imperfect: "Fix login issue for SSO users"
• Description exists and explains the change, even briefly
• Test plan references Jira task with testing details

Format

- ❓ **QUESTION**: PR title could be more specific
  - Suggested: "Fix null check in UserService.getProfile"

Summary Format

## 🤖 Bitwarden Claude Code Review
 

No Verdict

Use this form instead of the template above — not as a third value inside it — when nothing was reviewed — the diff could not be produced, it came back empty, the pull request could not be identified, or any other reason the review did not happen. Emit no APPROVE or REQUEST CHANGES; a verdict over an unreviewed diff is worse than none.

It goes to the same destination the mode you are in would have used for a normal summary, by the routing in Context Detection above and the per-mode steps in Output Execution below — including the marker append in agent mode, without which the next run cannot find the sticky comment and duplicates it. Nothing about this form changes where output lands — a stop that writes nowhere leaves a placeholder comment reading as a review that found nothing, which is the outcome the form exists to prevent.

The reason goes in the assessment-sentence position, outside <details>, for the same reason the coverage note does — a reason nobody expands is a reason nobody reads. There is no <details> block at all here, because there are no findings to put in one, and no coverage note either: this form already says the whole review did not happen.

Not Covered

A coverage note is not a finding: it has no file:line to cite and it says what was not reviewed, which is the opposite of the assessment sentences above it. So it gets its own line, outside the <details> block, where an approval cannot bury it.

Render it when a review step the diff called for did not run and nothing else in this review covered its ground — this path cannot reach that reviewer at all, or the reviewer errored or returned nothing usable. An optional enrichment skill that was merely unavailable does not qualify: those fall back to your existing review knowledge, so the ground was still covered. Name the files that went unreviewed, why, and which other path covers them. Give the reason that actually applies on your path — do not assert an install state you cannot check, and do not point at a remedy the stated reason would also block.

Example, the bitwarden-code-reviewer path, where Task is unavailable so plugin-dev:skill-reviewer cannot be launched whatever is installed:

Omit the line entirely when every review step the diff called for ran. An approval that hides what went unreviewed reads as a pass on it.

Dependency Changes Table

When the PR diff includes dependency manifest file changes, add a Dependency Changes subsection inside the <details> block, after the findings list and before the optional PR Metadata Assessment.

Only render this table when there are meaningful version changes — not for lock file-only churn with no manifest changes.

Bold the word "major" for major version bumps. Mark new additions as "New (version)" and removals as "Removed".

Findings in Details Section

Ordering: Group findings by severity in this exact order:

1. ❌ : CRITICAL
2. ⚠️ : IMPORTANT
3. ♻️ : DEBT
4. 🎨 : SUGGESTED
5. ❓ : QUESTION

Omit empty categories entirely.

Format per finding:

Example:

Output Execution

Agent Mode (Sticky Comment)

When sticky comment context is provided in the prompt (comment ID + marker):

1. Write the summary to /tmp/review-summary.md using the Write tool
2. Append \n\n<!-- bitwarden-code-review --> at the end of the file content
3. Do NOT use mcp__github_comment__update_claude_comment
4. Do NOT use gh pr comment or gh api

The workflow post-step will read this file and update the placeholder comment automatically.

GitHub Actions (Tag Mode)

Local