Reply to PR review comments after addressing them. Resolves conversations where changes were made. Uses polite tone for humans, terse factual responses for AI bots.
Installs into .claude/skills of the current project.
Are you the author of Reply Comments Flurdy Agent Skills 2?
Add the live security badge to your README. It updates with every re-scan.
[](https://www.skillsdirectory.com/skills/majiayu000-reply-comments-flurdy-agent-skills-2)
---
name: reply-comments
description: Reply to PR review comments after addressing them. Resolves conversations where changes were made. Uses polite tone for humans, terse factual responses for AI bots.
---
# Reply to Review Comments
Reply to PR review comments after addressing the feedback. Use this after `/review-comments` to close the feedback loop.
## Usage
```
/reply-comments
/reply-comments 123 # Specific PR number
```
## Instructions
### 1. Find the PR
If no PR number provided, get it from the current branch:
```bash
gh pr view --json number,url,title,headRepositoryOwner,headRepository \
--jq '{number, url, title, owner: .headRepositoryOwner.login, repo: .headRepository.name}'
```
### 2. Fetch Review Comments
Get all review comments with their thread/resolution status:
```bash
# Get inline code review comments
gh api repos/{owner}/{repo}/pulls/{pr_number}/comments \
--jq '.[] | {id, path, line, body, user: .user.login, in_reply_to_id, created_at}'
```
### 3. Get Recent Commits
Check what was recently committed to understand which comments were addressed:
```bash
# Get recent commit messages and changed files
git log --oneline -10
git diff HEAD~1 --name-only
```
### 4. Identify Addressed Comments
For each review comment, determine if it was addressed by:
- Checking if the file/line was modified in recent commits
- Matching commit messages to comment content (e.g., "address review feedback")
- Looking for code changes that match suggested fixes
### 5. Compose Replies
Use different tones based on the reviewer:
**AI Bots** (amazon-q-developer[bot], copilot[bot], github-actions[bot], etc.):
- Terse, factual responses
- Just state what was done
- Examples:
- "Fixed."
- "Done."
- "Changed to use `const`."
- "Added null check."
- "Not applicable - already handled by X."
**Human Reviewers** (anyone without [bot] suffix):
- Short but polite responses
- Acknowledge their feedback
- Examples:
- "Good catch, fixed!"
- "Thanks - updated."
- "Done, good suggestion."
- "Makes sense, changed it."
- "Addressed in latest commit."
For comments NOT addressed (intentionally skipped):
- "Keeping as-is because {brief reason}."
- "Intentional - {brief explanation}."
### 6. Post Replies
Reply to each comment:
```bash
gh api repos/{owner}/{repo}/pulls/{pr_number}/comments/{comment_id}/replies \
-f body="{reply_text}"
```
### 7. Resolve Threads (if addressed)
Use GraphQL to resolve review threads where changes were made:
```bash
# First, get the thread ID for the comment
gh api graphql -f query='
query($owner: String!, $repo: String!, $pr: Int!) {
repository(owner: $owner, name: $repo) {
pullRequest(number: $pr) {
reviewThreads(first: 100) {
nodes {
id
isResolved
comments(first: 1) {
nodes {
databaseId
body
}
}
}
}
}
}
}
' -f owner="{owner}" -f repo="{repo}" -F pr={pr_number}
# Then resolve each thread that was addressed
gh api graphql -f query='
mutation($threadId: ID!) {
resolveReviewThread(input: {threadId: $threadId}) {
thread {
isResolved
}
}
}
' -f threadId="{thread_id}"
```
### 8. Summary
Report what was done:
```
Replied to 6 comments:
- amazon-q-developer[bot]: 3 (all resolved)
- @username: 2 (2 resolved, 1 kept as-is)
- copilot[bot]: 1 (resolved)
```