The News, Explained

On November 9, 2026, GitHub announced a public preview of secret scanning for GitHub Actions workflow logs. Actions automates work such as code checks and deployments; logs record its output. The new scanning surface covers exposed credentials such as API keys and tokens that can grant access to other services. Source

Scanning happens after a workflow run completes. Findings appear in the repository’s secret scanning alert list alongside alerts from code and other content, with a link to the relevant log. Validity checks run for supported secret types to check whether credentials remain active. Generated alerts survive deletion of the logs. Source

The preview is for enterprises on GitHub Enterprise Cloud with GitHub Secret Protection or GitHub Advanced Security, including data residency on GHE.com. It is off by default. Enterprise owners must open Policies, then Advanced Security in enterprise settings and set Secret scanning for Actions workflow logs to Enabled in the All repositories dropdown. The policy applies only to repositories with secret scanning enabled under those licenses. Source

There are two boundaries to coverage. Scanning covers new runs that complete after the feature is enabled, with no backfill of existing logs. It detects only provider patterns, the credential formats associated with service providers. Generic patterns, custom patterns, AI-detected secrets, and values already masked by Actions are excluded from this log scanning preview. Source

Teams therefore need to distinguish historical logs from new runs when enabling the feature. For example, a completed run’s records can remain available while falling outside this preview’s backfill coverage. Even when a new run produces findings, alerts do not generate notifications during the public preview. The repository’s secret scanning alert list must be reviewed directly. Source

The documentation says one alert can group locations of the same secret across multiple workflow runs or jobs. Each location links to the workflow file and the log line where the secret was printed; the alert has no inline preview of the log content. This distinction matters when interpreting alert counts and the number of runs that exposed a credential. Source

When an exposed credential is found, GitHub’s documentation directs users to rotate it immediately: issue a replacement and revoke the old credential. Removing a log record and withdrawing a credential’s access are separate actions. This scanning operates after a run, helping teams detect exposure and respond. Source

OYOPICK’s Take

We welcome the extension of detection to automated runs’ output. Grouping repeated appearances of a secret into one alert could help teams trace exposure paths as well as count findings. With historical logs and several pattern types outside coverage, however, an absence of alerts needs to be interpreted within the scope this preview actually checks.

We think the preview’s practical value depends on assigning responsibility for reviewing the alert list and rotating credentials after enabling it. Post-run scanning and the lack of notifications make those responsibilities particularly relevant. We hope notification paths and detection coverage develop further, reducing the burden of responding to exposure while preserving the convenience of automation.