pr-review-automation

name	pr-review-automation
description	Automate PR review workflows: fetch and apply Copilot review suggestions, check/fix broken guard checks (commitlint, PR validation, branch policies), and apply test recommendations.

PR Review & Guard Check Automation

Overview

This skill automates the common workflow of:

1. Fetching AI code review — Pull Copilot review suggestions from active/open PRs
2. Applying suggestions — Implement suggested changes with proper testing
3. Fixing broken guards — Diagnose and fix commitlint, branch validation, PR title format, and other checks
4. Validating changes — Re-run guards to ensure everything passes

Typical workflow:

gh pr view 161 --json comments  [fetch Copilot review]
  ↓
[Analyze suggestions + implement changes]
  ↓
git commit + git push  [push changes]
  ↓
gh pr view 161 --json statusCheckRollup  [verify all checks pass]

---

Key Operations

1. Fetch Copilot Code Review Comments

Goal: Retrieve AI-generated review suggestions from open PR

Command:

gh pr view <NUMBER> --json comments --jq '.comments[] | select(.author.login | contains("copilot")) | {body, url, state}'

Also check: Review comments (line-level suggestions), not just PR-level comments

gh api repos/{owner}/{repo}/pulls/{number}/comments --jq '.[] | select(.user.login | contains("copilot"))'

Output to check:

• Comment author is copilot-* bot
• Comment contains // suggestion: or similar marker
• Line references specific code location
• Suggestion includes rationale

---

2. Detect & Fix Broken Guard Checks

Common guard checks that fail:

Guard	Failure Message	Fix
Commitlint	"type must be one of..."	Update commit message type (feat/fix/docs/etc)
Validate PR Title	"PR title must match pattern"	Update PR title format (type(scope): subject)
Require Issue Reference	"No issue linked"	Link issue in PR description or commit
Branch Naming	"Branch must follow pattern"	Rename branch (feature/, fix/, docs/, etc)
Prevent Direct Push	"Branch is protected"	Use PR instead of direct push

Detection strategy:

# Get failed status checks
gh pr view <NUMBER> --json statusCheckRollup --jq '.statusCheckRollup[] | select(.conclusion != "SUCCESS" and .conclusion != "SKIPPED")'

# Or get logs (if available)
gh run list --status failure --branch <branch> --limit 1 --json databaseId,conclusion

Fix strategies:

Commitlint Fix

# Check what failed
git log --oneline HEAD~3..HEAD | head -1  # See last commit

# Amend if type is wrong
git commit --amend --message "fix(scope): new message"  # Change type: style→fix, wrong scope, etc
git push --force-with-lease

PR Title Format

# Update PR title (requires PR number)
gh pr edit <NUMBER> --title "feat(scope): description here"

Link Issue

# Add issue to PR description or use CLI
gh pr edit <NUMBER> --add-assignee @me
# Or edit description manually to add "Closes #<issue>" or "Related to #<issue>"

---

3. Apply Copilot Suggestions

Workflow:

1. Parse suggestion from comment

   • Extract code location (file, line range)
   • Extract intended change
   • Identify type (assert strengthening, mock improvement, documentation, etc)

2. Apply programmatically or manually

   # For straightforward suggestions: edit file directly
   git checkout <BRANCH>
   # Make changes
   git add .
   git commit -m "review(scope): apply Copilot suggestion — [brief description]"
   git push
   

3. Test & validate

   pytest tests/...  # Run affected tests
   ruff check .      # Lint
   mypy .            # Type check
   

4. Verify guard checks pass

   gh pr view <NUMBER> --json statusCheckRollup --jq '.statusCheckRollup[] | select(.conclusion != "SUCCESS" and .conclusion != "SKIPPED")'
   # Should return empty if all pass
   

---

Execution Checklist

Before applying suggestions:

• Fetch all Copilot comments from PR
• Verify each suggestion makes sense in context
• Identify dependencies between suggestions (apply in order)
• Check if test modifications require coverage verification

During implementation:

• Use atomic commits (one suggestion = one commit when practical)
• Commit message: review(scope): apply Copilot suggestion — <brief desc>
• Follow project commitlint rules (type(scope): subject)
• Run local tests before pushing

After pushing:

• Wait for CI to complete
• Check for remaining guard failures
• If failures detected, diagnose and fix immediately
• Re-verify all guards pass

For guard check failures:

• Identify which guard failed (commitlint, title format, branch name, etc)
• Read failure message carefully
• Fix at source (commit message, PR title, branch name, issue link)
• Verify fix before next push (local validation where possible)
• Force-push if needed (git push --force-with-lease)

---

Example: Complete Workflow

Scenario: PR #161 with 3 Copilot suggestions + commitlint failure

Step 1: Fetch suggestions

gh pr view 161 --json comments,reviews \
  --jq '.reviews[] | select(.author.login | contains("copilot"))'

Step 2: Identify guard failure

gh pr view 161 --json statusCheckRollup \
  --jq '.statusCheckRollup[] | select(.conclusion != "SUCCESS" and .conclusion != "SKIPPED")'
# Output: Commitlint failed

Step 3: Fix commitlint

git log --oneline -1  # See what's wrong
git commit --amend --message "test(sagaz): apply Copilot suggestion — strengthen assertions"
git push --force-with-lease

Step 4: Apply suggestions (3 separate commits)

# Suggestion 1: Add type check to test_redis_storage_no_url_uses_default
git add tests/...
git commit -m "review(tests): add Redis type validation to default URL test"

# Suggestion 2: Add assertion to test_redis_storage_empty_url_skips_default  
git add tests/...
git commit -m "review(tests): assert InMemorySagaStorage on empty URL fallback"

# Suggestion 3: Fix comment accuracy
git add tests/...
git commit -m "review(tests): clarify _parse_storage_url behavior in test docstring"

git push

Step 5: Verify all checks pass

# Wait for CI, then verify
gh pr view 161 --json statusCheckRollup | grep -c "SUCCESS"
# Should show all checks passing

---

Anti-Patterns to Avoid

❌ Don't:

• Apply suggestions without reading them carefully (may be incorrect context)
• Mix multiple unrelated suggestions in one commit
• Ignore commitlint failures (signal of bigger PR mismatch)
• Force-push without --force-with-lease (risky in team settings)
• Apply test suggestions without running full test suite

✅ Do:

• Review each Copilot suggestion in context before applying
• Use atomic, well-described commits for traceability
• Fix guard checks immediately when they fail
• Always test locally before pushing
• Use --force-with-lease when amending committed work

---

Related PR Policies

This skill complements the development-policy skill:

• Commitlint types must be from approved list: feat|fix|docs|refactor|test|perf|build|ci|chore|revert
• PR titles must match type(scope): subject format
• Scope must be non-empty and match approved project scopes
• Subject must not use sentence case or end with period
• Issues must be referenced in PR description or commit message

See .github/skills/development-policy/SKILL.md for full policy.

---

Quick Reference Commands

# Fetch all Copilot review comments on PR
gh pr view <NUMBER> --json comments --jq '.comments[] | select(.author.login | contains("copilot"))'

# Check failed guard checks
gh pr view <NUMBER> --json statusCheckRollup --jq '.statusCheckRollup[] | select(.conclusion != "SUCCESS" and .conclusion != "SKIPPED")'

# See commit history (for checking commitlint compliance)
git log --format="%h %s" | head -10

# Amend last commit (for fixing commitlint/PR title issues)
git commit --amend --message "fix(scope): corrected message"

# Safe force push (protects against concurrent changes)
git push --force-with-lease

# Run local tests before pushing
pytest tests/ -x  # Stop on first failure
ruff check .      # Lint