GitHub Team now runs security assessments automatically, but coverage still needs a check

GitHub said on November 18 that free security risk assessments now run automatically for organizations using GitHub Team. Organization administrators and security managers can review the findings to identify exposed credentials and code vulnerabilities and decide what to address first.
For a small software team, the useful change is that an initial assessment no longer depends entirely on someone remembering to start it. What follows still depends on someone reading the result. An automatically produced report is not evidence that every repository was examined or that a vulnerability has been fixed.
Check the boundaries of the report
GitHub's documentation describes a code assessment covering up to 20 repositories. Its default selection uses recent commit activity in private and internal repositories, and only repositories with a supported programming language qualify. The documented self-service workflow allows the selection to be changed before a scan.
That distinction matters when interpreting the new automatic behavior. The announcement does not say that the entire organization receives continuous, comprehensive coverage. Teams should inspect the repositories actually listed in their report rather than apply the result to code that was never included.
The documentation also describes a one-hour scan timeout. A repository can appear in the report if at least one language was scanned successfully, even when another language failed. A clean-looking aggregate therefore deserves a second look at failures and omissions.
Secrets and code have separate results
The company presents code and secret assessments in separate tabs. The former concerns vulnerabilities in software; the latter concerns exposed credentials. A reassuring result in one tab does not answer the question asked by the other. GitHub documents a 90-day interval for rerunning each assessment.
The >November 18 announcement (opens in a new tab) describes the assessments as free. The code-assessment documentation also says that the scan does not charge Code Security licenses or the Actions minutes it uses. That is a statement about the assessment, not a promise that every subsequent security service is included.
Turn findings into assigned work
A practical first review should record which repositories were covered, which checks failed and who owns each actionable finding. If the report identifies a credential, investigate its status and handle revocation through the issuing service; deleting visible code alone is not a substitute for addressing access.
For code findings, review the proposed repair and test the affected behavior before treating the work as finished. Automation can make the first warning arrive sooner. The team's own records must still show what was investigated, changed and checked.


