pr-review-resolver
Skill2 starsChanged 10 months agoMITView on GitHub
Automatically monitor, evaluate, and resolve CodeRabbit and Qodo PR review comments. Waits for review bots to comment, evaluates suggestions intelligently, applies fixes, and marks comments as resolved. Designed to run in ralph-loop for continuous operation.
- Pipes a download into a shell
- Reads credentials
- Commits and pushes
2 other projects use a version of this.
Adopted by
Most recently taken up by
- realityforge/skillstheir own version, first seen Oct 20, 2026
- gerivdb/SKILLStheir own version, first seen Aug 7, 2026
Other versions2 projects use a version of this.
Word for word · 0
None found.
Edited · 0
None found.
Wrote their own version · 2
- gerivdb/SKILLS★ 1same name
- realityforge/skills★ 0same name
How matches are found. Word for word: the exact same text. Edited: shares passages with this file (“99% same” means 99% of its five-word phrases also appear here; 10% or more counts). Their own version: similar meaning, or, weakest, the same name. We can’t tell which project had it first. Projects are counted among the 200 most-starred candidates, and mentions are links to the repository we found, up to 50 per site, so both counts are a floor.
In this repository17 other files it gives its agents.
- react-best-practices/AGENTS.mdAGENTS.md
- agent-browserSkill
- claudeceptionSkill
- devlog-harvesterSkill
- humanizerSkill
- knowledge-forgeSkill
- loop-maestroSkill
- notebooklmSkill
- obsidian-bridgeSkill
- portfolio-forgeSkill
- project-analyzerSkill
- vercel-react-best-practicesSkill
- vaultage-captureSkill
- vaultage-dailySkill
- vaultage-inboxSkill
- vaultage-searchSkill
- web-design-guidelinesSkill
The file
We split this skill into the rules it gives an agent
6 rules. About 0 are common, 6 are distinctive, and 0 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈6
Distinctive · estimate
This author's own choices. Worth a look.
0
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Distinctive: line 103
Check if multiple suggestions target same lines
- Distinctive: line 138
If running in ralph-loop mode:
- Distinctive: line 140
Wait for new review cycle (2-5 minutes)
- Distinctive: line 141
Check for new comments
- Distinctive: line 305
Wait for CodeRabbit/Qodo to re-review
- Distinctive: line 325
Wait for rate limit reset or use a token with higher limits.
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
PR Review Resolver
Automated workflow for handling CodeRabbit and Qodo PR review comments across all active repositories.
Overview
This skill monitors PRs for review bot comments, evaluates suggestions, applies fixes intelligently, and resolves comments. It can run continuously in a ralph-loop until all issues are addressed.
Quick Start
# Resolve comments on a specific PR
/pr-resolve owner/repo#123
# Resolve comments on current branch's PR
/pr-resolve
# Run in continuous mode (ralph-loop)
/ralph-loop "/pr-resolve owner/repo#123 --loop"Supported Review Bots
| Bot | Detection | Comment Format |
| CodeRabbit | coderabbitai[bot] | Severity markers, diff blocks, committable suggestions |
| Qodo | qodo-code-review[bot] | Suggestion blocks with direct replacements |
Workflow
Phase 1: Discovery
# Find open PRs with pending review comments
gh pr list --state open --json number,title,author,headRefName
# Get review comments for a PR
gh api repos/{owner}/{repo}/pulls/{pr}/commentsPhase 2: Parse Comments
CodeRabbit Format:
_⚠️ Potential issue_ | _🔴 Critical_
**Issue title**
Description...
<details>
<summary>🔧 Proposed fix</summary>
```diff
-old code
+new code</details>
```Qodo Format:
**Suggestion:** Title
```suggestion
replacement code here
### Phase 3: Evaluate Each Issue
For each comment, evaluate:
1. **Validity Check**
- Is the suggestion syntactically correct?
- Does the file/line still exist?
- Has it already been addressed?
2. **Impact Assessment**
- Security issues (🔴 Critical) → High priority
- Logic bugs (🟠 Major) → Medium priority
- Style/lint (🟡 Minor) → Lower priority
3. **Conflict Detection**
- Check if multiple suggestions target same lines
- Determine which takes precedence
### Phase 4: Apply Fixes
```bash
# Read the target file
cat path/to/file.py
# Apply the fix (evaluated, not blind)
# For diff blocks: parse and apply hunks
# For suggestion blocks: replace exact lines
# Commit with reference
git add path/to/file.py
git commit -m "fix: address CodeRabbit review - <issue summary>
Resolves review comment: <comment_url>"Phase 5: Resolve Comments
# Reply to comment indicating fix
gh api repos/{owner}/{repo}/pulls/{pr}/comments/{comment_id}/replies \
-f body="✅ Fixed in commit $(git rev-parse --short HEAD)
Applied the suggested change. The issue has been addressed."
# Push changes
git pushPhase 6: Loop Until Done
If running in ralph-loop mode:
- Push changes
- Wait for new review cycle (2-5 minutes)
- Check for new comments
- Repeat until no unresolved comments remain
- Signal completion with
<promise>DONE</promise>
Comment Parsing Scripts
Extract CodeRabbit Issues
import re
import json
def parse_coderabbit_comment(body: str) -> dict:
"""Parse CodeRabbit comment into structured data."""
result = {
"severity": None,
"title": None,
"description": None,
"diff": None,
"suggestion": None
}
# Extract severity
if "🔴 Critical" in body:
result["severity"] = "critical"
elif "🟠 Major" in body:
result["severity"] = "major"
elif "🟡 Minor" in body:
result["severity"] = "minor"
# Extract diff block
diff_match = re.search(r'```diff\n(.*?)```', body, re.DOTALL)
if diff_match:
result["diff"] = diff_match.group(1)
# Extract committable suggestion
suggestion_match = re.search(
r'📝 Committable suggestion.*?```\w*\n(.*?)```',
body, re.DOTALL
)
if suggestion_match:
result["suggestion"] = suggestion_match.group(1)
return resultExtract Qodo Issues
def parse_qodo_comment(body: str) -> dict:
"""Parse Qodo comment into structured data."""
result = {
"title": None,
"suggestion": None
}
# Extract title
title_match = re.search(r'\*\*Suggestion:\*\*\s*(.+)', body)
if title_match:
result["title"] = title_match.group(1).strip()
# Extract suggestion block
suggestion_match = re.search(r'```suggestion\n(.*?)```', body, re.DOTALL)
if suggestion_match:
result["suggestion"] = suggestion_match.group(1)
return resultEvaluation Criteria
When to Apply a Suggestion
✅ Apply when:
- Suggestion is syntactically valid
- Target lines exist and match expected content
- Change doesn't introduce new issues
- Security fix (always prioritize)
- Clear bug fix with obvious improvement
⚠️ Evaluate carefully when:
- Suggestion conflicts with project conventions
- Multiple suggestions affect same code
- Change has broader implications
- Refactoring suggestion (may need context)
❌ Skip when:
- Target code has already changed
- Suggestion would break functionality
- False positive detection
- Requires human decision (architectural)
Priority Order
- 🔴 Critical - Security vulnerabilities, data loss risks
- 🟠 Major - Logic bugs, error handling issues
- 🟡 Minor - Style, lint, documentation
Environment Variables
| Variable | Description | Default |
GITHUB_TOKEN | GitHub personal access token | Required |
PR_RESOLVE_AUTO_PUSH | Auto-push after fixes | true |
PR_RESOLVE_WAIT_TIME | Seconds to wait for new reviews | 300 |
PR_RESOLVE_MAX_ITERATIONS | Max loop iterations | 10 |
Example Session
User: /pr-resolve chipoto69/ATLAS-CORPUS#4 --loop
Agent: 🔍 Fetching PR #4 from chipoto69/ATLAS-CORPUS...
Found 30 review comments:
- coderabbitai[bot]: 28 comments (5 critical, 12 major, 11 minor)
- qodo-code-review[bot]: 2 comments
📋 Processing by priority...
[1/30] 🔴 CRITICAL: launchd plist will fail due to unexpanded $HOME
File: corpus/avatars/templates/candysoul/bootstrap/README.md
Issue: Heredoc uses single quotes preventing variable expansion
Evaluating fix...
✅ Fix is valid and addresses the issue
Applying: Changed 'EOF' to EOF for variable expansion
Committed: abc1234
[2/30] 🔴 CRITICAL: Avoid "curl | bash" in Quick Start
File: corpus/avatars/templates/candysoul/bootstrap/README.md
Issue: Encourages execution of unreviewed remote code
Evaluating fix...
✅ Suggested download-inspect-run flow is safer
Applying: Added safer installation instructions
Committed: def5678
... (continues through all issues)
📤 Pushing changes...
⏳ Waiting 5 minutes for new review cycle...
🔄 Checking for new comments...
No new unresolved comments found.
<promise>DONE</promise>
✅ All review comments resolved!
- Fixed: 28 issues
- Skipped: 2 (already addressed)
- Commits: 15Integration with ralph-loop
This skill is designed to work with oh-my-opencode's ralph-loop:
/ralph-loop "/pr-resolve chipoto69/ATLAS-CORPUS#4"The loop will:
- Process all current comments
- Push fixes
- Wait for CodeRabbit/Qodo to re-review
- Address any new comments
- Repeat until
<promise>DONE</promise>
Files Modified
The skill creates no persistent files. All state is managed through:
- Git commits (fix history)
- GitHub API (comment status)
- PR conversation thread (resolution replies)
Troubleshooting
"Comment already resolved"
The comment may have been addressed in a previous run. Check git log for related commits.
"Cannot apply diff - context mismatch"
The target code has changed since the review. Re-fetch the file and manually verify the fix.
"Rate limited by GitHub API"
Wait for rate limit reset or use a token with higher limits.
"Suggestion introduces syntax error"
Skip the suggestion and flag for human review. Log the issue.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

