Use these questions to examine how a real change, vulnerability, or exception moves through your security process. Start with an application you know, then consider others whose exposure, technology, size, or release pattern could produce a different answer.
Look at the relevant settings, pipeline results, findings, and decisions. Ask the people involved when a handoff isn't clear, and compare their experience with what the records show. For a gap you choose to address, identify the risk, who can improve it, and the result you'll check afterward.
Checks that cover the risks you face
- Do checks act early enough to prevent problems where prevention is possible? Consider what a safe test secret or insecure change would encounter before commit, during review, and after merge.
- Do the chosen scanners cover the application's actual technology and delivery path? Check languages, package sources, containers, and running targets. An enabled scanner that doesn't examine the relevant code or artifact leaves a gap.
- Have teams reviewed findings from new scans on representative projects and been able to act on them? Include different repository sizes, network restrictions, and runner needs before extending configuration to more teams.
- Can you tell a clean scan from a missing, skipped, or failed scan? Inspect job completion and report contents, including any child pipeline results the workflow relies on.
- Does another control address a problem the first check might miss? Consider human review, dependency checks, artifact verification, and response procedures as well as repeated scanning.
- Can the people handling findings keep up with the rollout? Look at time to first review, unresolved ownership, false positives, and runner demand before broadening coverage.
When these questions reveal missing or unreliable scans, Security scanning that covers what you ship helps you choose coverage and work through the first results with application teams.
Feedback developers can use
- Does a developer receive enough information before merge to act? Look for the affected file or package, the nature and severity of the problem, and guidance on how to fix it in the merge request or the team's report-review process.1
- Can a developer run appropriate checks and find a remediation playbook without waiting for security? Try the path for a common finding and note where help is still needed.
- Does feedback distinguish a new finding from a problem already present in the code? Check the source and comparison reports before relying on the merge request result.
- Can someone get an informed security answer within the time their work needs? Follow a recent question or exception request through to the person who could decide it.
- Has a new approval policy been tried against real work before becoming fully blocking? Where warn mode is used,2 review violations, bypass reasons, false positives, and delay. Keep existing mandatory protections in view during the trial.
- Have recurring questions or developer difficulties led to better rules or guidance? Compare developers' experience with the policy owners' expectations and check what changed.
Required controls that remain in effect
- Would a required scan still run if a project changed or removed its local include? Shared configuration helps adoption; check the separate enforcement mechanism for mandatory coverage.
- Can you confirm that the policy applies to the intended projects and refs? Compare policy links and scope with actual jobs and reports, including after a policy change has synchronized.
- Are policy changes reviewed independently of the application work they affect? Check who can propose, approve, merge, and delete policy changes, rather than relying only on the repository's name.
- Can an application role weaken or remove its own required control? Examine inherited permissions, policy-project membership, link and scope permissions, and non-human access. Try the relevant action with an ordinary delivery role.
- Are policy files and their supporting scanner configuration protected from unintended changes? Review default-branch protection and change approval alongside who can edit the CI/CD configuration the policy runs.
- Can delivery teams read what they need and propose corrections without unrestricted policy access? For pipeline execution policies, check the required read access to referenced configuration too.
- Can an urgent exception be reviewed promptly and end when it's no longer needed? Look for an authorized approver, reason, affected scope, expiry or review date, and follow-up work.
For gaps in enforcement or policy ownership, Security policy from observation to enforcement explains how to introduce a requirement, examine its effect, and extend it to more projects.
Findings that reach a supported decision and fix
- Can you follow a current finding to its owner and latest decision? Check the scan, investigation, linked fix, dismissal reason, or accepted-risk review.
- Can new urgent findings get attention while the initial backlog is being triaged? Look for a route that doesn't leave a newly exposed credential waiting behind older low-impact findings.
- Does prioritization reflect application impact and exposure as well as severity? Check how reachability, exploit likelihood, data sensitivity, and compensating controls influenced a recent decision.
- Are uncertain findings investigated before being dismissed as false positives? The record should explain what was checked and why the finding doesn't apply.
- Does closure reflect the code and versions people actually use? Compare the fix, a successful scan, the relevant branch or tracked ref, and any deployment or backport still needed.
- If a finding is no longer detected or appears again, does someone check why? Changes in scan coverage can resemble a fix, and a recurrence may reveal an incomplete one.
- For an exposed secret, was the credential revoked or rotated as well as removed from code? Check the action that ended the old access.
- Are accepted risks reconsidered when their assumptions change? Review expired exceptions and changes to application exposure or compensating controls.
- Have repeated weaknesses led to safer shared patterns or scanner tuning? Look for follow-up beyond closing individual tickets, and review consequential status changes made by automation.
If findings get lost between teams or close without a confirmed fix, Vulnerability management from finding to closure helps you establish the ownership and follow-through that are missing.
Releases and records you can investigate
- Can you connect an SBOM to the release it describes? Check direct and indirect dependencies, supplier components, and whether the data is current enough to locate a newly affected release.
- Can you verify the origin and digest of the artifact being delivered? Compare it with the expected build and provenance. An approval based on a scanner report doesn't verify that report's authenticity.
- Are the scan, risk decision, fix, exception, and deployment records available together when needed? Follow a past finding far enough to explain both the action and its reason.
- Does a compliance status result cover the requirements and period you're reviewing? Check the configured controls and scan time rather than assuming a current result describes the whole period.3
- Can you retrieve the access and policy changes needed for an investigation? Review the relevant audit events and their query or export limits.4
- Will required reports and artifacts remain available for later review? Check retention and export completeness, including project and ref scope and results produced in child pipelines. Arrange additional storage where the available records won't meet the need.
When the gap is between a reviewed build and what reaches users, deployment and recovery planning explains how to keep release identity, promotion, and recovery connected.
Further Reading
Notes
-
Integrated merge request security reports and the vulnerability report are available in Ultimate. Teams using scanner artifacts directly need another way to review and track findings. ↩
-
Security policies, including policy projects and approval-policy warn mode, are available in Ultimate. ↩
-
The compliance status report is available in Ultimate and reports checks against configured framework requirements. ↩
-
Group and project audit events are available in Premium and Ultimate. Successful sign-in events are available on all tiers. ↩
- Interpret security scanning results when findings are missing, unclear, or difficult to review.
- Manage policy projects and permissions when mandatory controls can be changed by the work they govern.
- Pilot merge request approval policies when evaluating a new rule's effect on delivery.
- Work through the vulnerability report when triage, ownership, or closure needs attention.
- Protect the pipeline and software supply chain when build inputs, reports, or released artifacts need stronger verification.
- Interpret compliance status and retrieve audit events when preparing records for review.