How do I check my AI Code Insights commits?

Use this guide to review your recent commits in DX and check their AI code percentage. Checking a few commits helps your organization confirm that AI attribution looks accurate before expanding the rollout of AI Code Insights.

This guide is for users of the DX AI Code Insights daemon, which powers features such as AI code percentage and Agent Experience Score.

Before you begin

You need:

• Access to DX through your organization’s workspace.
• AI Code Insights installed on the machine where you write and commit code.
• Recent commits in repositories monitored by DX, preferably commits where you used an AI coding tool.

Check your commits

1. Open the AI code percentage report in DX.
2. Find your team in the breakdown table below the chart.
3. Click the Overall average percentage for your team to view the underlying commits.
4. Find 2–3 recent commits where you used a supported AI coding tool.
5. Review the AI percentage shown for each commit.
6. If a commit looks inaccurate, hover over its row and click the pencil icon next to AI code %.
7. Enter the percentage you believe is correct, from 0 to 100, and click Save.

Correct an AI code percentage

You can edit commits linked to your DX user. Workspace admins and Data Cloud admins can edit any commit they can access. If the pencil icon is missing on your own commit, ask an admin to check that the commit’s Git email is linked to your DX user.

Saving sets a manual override for that commit. The row shows the corrected percentage and an Edited indicator. The AI code percentage report uses the override in its weighted calculations. You can click the pencil icon again to change the value.

A measured 0% result means DX recorded no AI-attributed changes in the captured denominator. It does not prove that no AI was used; unsupported tools, pasted code, missed events, and work before installation can lack an AI signal. A dash indicates unmeasured attribution. An Edited value reflects a manual override.