• Sources: primary
  • Summary: A new copilot_code_review object on repos-1-day rows reports total_comments_evaluated and total_comments_addressed, broken down by severity, review tier, comment type and customization such as custom instructions, skills or MCP servers. It also reports total_merged_no_human_change_request_after_review. Comments are counted on the day their pull request merges, and comments on bot-authored pull requests are usually not evaluated.
  • Why it matters: Engineering leads can now measure whether AI pre-review feedback is acted on per repository and whether custom instructions or skills change that.