Comment on a pull request

Workflows•3 min read|
|

Use comments when you want the author to make a change, explain a decision, or confirm accuracy.

Before you start

  • Open the PR in DraftView.
  • Confirm you have comment permission for that queue item.

Steps

Open the file in the visual surface, then highlight the sentence or block you want to discuss.

State what should change and why. Use concrete language that the author can act on.

Submit your comments from the review toolbar.

DraftView posts your comments to the PR review thread in GitHub.

Format your comment

DraftView renders comments as GitHub-flavored Markdown, so your comment looks the same in DraftView and in GitHub. You can use code spans, code blocks, lists, task lists, tables, and links.

Comments from GitHub, including review bots, render the same way. DraftView keeps the inline HTML that GitHub allows, such as badges and collapsible <details> sections, and removes scripts, styles, and other unsafe markup.

Reply to existing threads

DraftView shows the PR's existing GitHub review threads next to the text they discuss. Select Reply to answer a thread, or Resolve to close it. To open a resolved thread again, select Reopen.

DraftView holds your replies and resolve actions on GitHub threads until you submit your review, then sends them to GitHub with the rest of your feedback.

Comment on deleted content

DraftView shows content that the PR removes as inline deletion blocks in the review surface, including files that the PR deletes entirely.

To object to a deletion:

  1. Find the deletion block at the place where the content was removed.
  2. Select Comment on the block. DraftView quotes the deleted text into your comment so the author sees the full context.
  3. Submit your review. DraftView anchors the comment to the deleted lines in the GitHub diff, so the thread appears on the removed content itself.

Write comments that resolve faster

  • Tie each comment to one issue.
  • Include expected behavior or exact wording when possible.
  • Ask one clear follow-up question if data is missing.