[FEATURE] Allow deleting my own resolved PR review comments on explicit request

Status Open
Maintainer reply None cached
Activity 0 comments · opened Sep 23, 2026

Preflight Checklist

  • [x] I have searched existing requests and this feature hasn't been requested yet
  • [x] This is a single feature request (not multiple features)

Problem Statement

Please let Claude Code hard-delete pull request review comments when I ask it to, at least when I am the only author of every comment in the thread and the PR is my own. Today it refuses unconditionally — it says permanently deleting messages is a fixed safety limit that no request can override — and instead generates a gh api command for me to run myself. Until a few days ago this request was simply fulfilled.

Proposed Solution

Proposed behaviour

  • Deleting comments that I authored, on a PR I own, on my explicit request, is allowed.
  • If irreversibility is the concern, gate it behind the same confirmation prompt used for other destructive operations, ideally with a summary ("delete 23 comments across 10 resolved threads, all authored by you").
  • Keep the refusal for comments authored by other people, and for anything not explicitly requested.

Alternative Solutions

_No response_

Priority

High - Significant impact on productivity

Feature Category

Other

Use Case Example

I use Claude Code as a review partner on my own PRs:

  1. I open a PR and leave review comments on it myself, as notes for Claude.
  2. I ask Claude to address them. Claude replies in each thread as it makes the changes.
  3. I check the result; when I'm happy with a thread, I resolve it.
  4. Before asking teammates for review, I ask Claude to remove all resolved conversations "with myself" — threads where I am the only participant, as both reviewer and (via Claude) replier.

Step 4 is what makes the loop worth it. By the time teammates look at the PR, those threads are pure noise: a dozen or more resolved conversations, all authored by me, that need no one's attention. Deleting them keeps the review focused so colleagues only see what actually needs their eyes. Minimizing/collapsing is not a substitute — the threads still count, still appear in the timeline, and reviewers still have to skip past them.

Additional Context

_No response_

View original on GitHub ↗