review pr ci

Arguments

• $ARGUMENTS: The PR number to review

Your task

Review the specified pull request and post a review to GitHub. This command is designed for CI/automated environments.

Step 1: Gather PR context

Get the PR details and diff:

gh pr view $ARGUMENTS
gh pr diff $ARGUMENTS

Step 2: Review the code changes

Use the Task tool with subagent_type: "code-reviewer" to perform a thorough code review of the changes. Pass the diff and changed files to the agent for analysis.

The code-reviewer agent will analyze:

• Code quality and best practices
• Potential bugs or issues
• Security concerns
• Performance considerations
• Documentation and comments

Step 3: Determine review outcome

Based on the code review findings, determine the appropriate review action:

• APPROVE (--approve): Code looks good, no significant issues found
• REQUEST_CHANGES (--request-changes): Critical issues that must be fixed before merging
• COMMENT (--comment): Suggestions or minor issues that don't block merging

Step 4: Post the review

Post the review to GitHub using:

gh pr review $ARGUMENTS --body "YOUR_REVIEW_BODY" --approve|--request-changes|--comment

Format your review body using this template with collapsible sections:

## PR Review

**Recommendation**: APPROVE | REQUEST_CHANGES | COMMENT

### Summary
[1-2 sentence overview of what this PR does]

<details>
<summary>Actionable Feedback (N items)</summary>

List specific items that need attention. Use checkboxes for trackable items:

- [ ] `file.py:42` - Description of issue or required change
- [ ] `notebook.ipynb` (in cell with `some_code = ...`) - Description
- [ ] General: Description of non-file-specific feedback

</details>

<details>
<summary>Detailed Review</summary>

### Code Quality
[Analysis of code patterns, readability, maintainability]

### Security
[Any security considerations or concerns]

### Suggestions
[Optional improvements that aren't blocking]

### Positive Notes
[What was done well - be specific]

</details>

Guidelines:

• Keep the summary and actionable feedback visible (outside collapsed sections)
• Put detailed analysis in the collapsed "Detailed Review" section to reduce noise
• Use checkboxes in actionable feedback so authors can track what they've addressed
• For Jupyter notebooks, reference code snippets instead of cell numbers (e.g., "in cell with data = pd.read_csv(...)")

Important: The gh pr review command produces no output on success. Only run this command once - do not retry if there is no output, as that indicates success.