Context

•OWNER: !gh repo view --json owner -q .owner.login
•REPO: !gh repo view --json name -q .name
•PR_NUMBER: !gh pr view --json number -q .number

Your task

Fetch unresolved review threads using this GraphQL query (substitute OWNER, REPO, PR_NUMBER from context above):

bash

gh api graphql -f query='
query($owner: String!, $repo: String!, $pr: Int!) {
  repository(owner: $owner, name: $repo) {
    pullRequest(number: $pr) {
      url
      title
      reviewThreads(first: 100) {
        nodes {
          isResolved
          comments(first: 10) {
            nodes {
              author { login }
              body
              path
              line
            }
          }
        }
      }
    }
  }
}' -f owner="OWNER" -f repo="REPO" -F pr=PR_NUMBER

Filter the results to only show threads where isResolved: false.

Create a markdown file named pr-{number}-comments.md (using the actual PR number) in the current directory with this format:

markdown

# PR Review Comments

> [PR Title](PR URL)

## Unresolved Comments

- [ ] **@author** `#fix-null-check` in `path/to/file.ext:line`

  > Add null check before accessing property

- [ ] **@author** `#rename-variable` in `path/to/file.ext:line`
  > Use more descriptive variable name

Generate unique descriptive tags based on the comment content (e.g., #fix-null-check, #add-error-handling, #rename-variable) so I can copy-paste them to reference specific comments. Summarize each comment concisely instead of quoting the full text. Group comments by file path if there are multiple in the same file. Use checkboxes so I can track progress.

If there are no unresolved comments, do not create the file. Just inform me that there are no unresolved comments.

After creating the file, let me know you're ready to address comments one by one. I'll reference them by their tag (e.g., #fix-null-check).