Skip to main content

AI Usage > Review suggestions accepted

Written by Tom Williams

Dataset: AI Usage

Entity: AI Usage Measurement

Field ID: review_suggestions_accepted

Type: Number

Description: The number of AI code review suggestions that were applied. Distinct from suggestions_accepted, which covers IDE completions.

Source: App

Transformation logic: N/A

App Mapping

GitHub Copilot

pull_requests.total_copilot_applied_suggestions, on repository and organization records

Anthropic Claude (coming soon)

N/A

Cursor (coming soon)

N/A

OpenAI Platform (coming soon)

N/A

OpenAI Codex (coming soon)

N/A - Codex reports engagement with review comments, never whether one was applied

Reporting Use Cases

The Review suggestions accepted field counts review comments that were actually applied, which is the only honest measure of whether AI review is useful.

  • Value of AI review: SUM(review_suggestions_accepted) / SUM(review_suggestions_offered) shows what share of AI review comments were acted on. A low ratio with a high volume is a signal to tune or disable automatic review.

  • GitHub Copilot only: Codex publishes replies and reactions rather than applications, and those are deliberately not mapped here: engagement is not acceptance, and treating it as such would make Codex's rate mean something different from Copilot's.

Did this answer your question?