Code scanning on pull requests

Understand how CodeQL and AI Scan find and help prevent vulnerabilities in pull requests across your organizations.

Overview

The "Code scanning on pull requests" report in security overview helps you understand how well code scanning is preventing vulnerabilities in pull requests in your organization or across organizations in your enterprise. You can view the entire dataset or filter for specific criteria, making it easy to identify repositories where you may need to take action to find and reduce security risks.

The report has separate CodeQL and AI Scan tabs.

Available metrics

CodeQL metrics

The overview shows you a summary of how many vulnerabilities prevented by CodeQL have been caught in pull requests. The metrics are only tracked for pull requests that have been merged into the default branches of repositories in your organizations.

You can also find more granular metrics, such as how many alerts were fixed with and without Copilot Autofix suggestions, how many were unresolved and merged, and how many were dismissed as false positive or risk accepted.

You can also view:

• The rules that are causing the most alerts, and how many alerts each rule is associated with.

• The number of alerts that were merged into the default branch without resolution, and the number of alerts dismissed as an acceptable risk.

• The number of alerts that were fixed with an accepted Copilot Autofix suggestion, displayed as a fraction of how many total Copilot Autofix suggestions were available.

• Remediation rates, in a graph showing the percentage of alerts that were remediated with an available Copilot Autofix suggestion, and the percentage of alerts that were remediated without a Copilot Autofix suggestion.

• Mean time to remediate, in a graph showing the average age of closed alerts that were remediated with an available Copilot Autofix suggestion, and the average age of closed alerts that were remediated without a Copilot Autofix suggestion.

AI Scan metrics

The AI Scan tab shows:

• Total findings: The number of AI Scan findings.
• Findings with Autofix suggestions: The number and percentage of findings that had a Copilot Autofix suggestion.
• AI Scan findings in pull requests: A chart that groups findings over time by Autofix suggested and Autofix not suggested.
• Outcomes of findings with an Autofix suggestion: The share of findings where a fix was applied or the finding was dismissed.
• Outcomes of findings without an Autofix suggestion: The share of findings where a fix was applied or the finding was dismissed.

The date range and filters that you select apply to all AI Scan metrics displayed in the report.

Visibility

You can see code scanning metrics for a repository if you have:

• The admin role for the repository
• A custom repository role with the "View code scanning alerts" fine-grained permissions for the repository
• Access to alerts for the repository

Next steps

To view the "Code scanning on pull requests" report, see Viewing code scanning metrics for pull requests.