Skepra / Web recovery & resilience
How do I scan npm dependencies for vulnerabilities?
For a WordPress theme build or another client web project that uses npm, start with the build the team actually ships. A dependency report can identify a next review; it cannot by itself certify the runner, the published website or a recovery outcome.
Discuss the next maintenance decisionIdentify the project and review boundary
Record the repository revision, workspace, Node and npm versions, lockfile and build or runner image associated with the client release. Ask which parts produce deployable assets and which are tooling used only during the build. Do not assume that a scan from a different working directory describes the deployed artifact.
The agency and build owner should agree whether the decision concerns a normal dependency update or a suspected environment incident.
Run a reporting check before selecting a repair
In an authorized project environment, plain npm audit requests a report of known vulnerabilities from the configured registry. npm documents that it sends dependency information; follow the project policy before using private package metadata with an external registry. npm audit --json provides machine-readable output.
Preserve the command, date, project revision and result together so the reviewer can tie findings to a particular dependency tree.
Read the result as a routing decision
For each material finding, identify the package, the dependency path, the published affected range, any documented fix and the owner of the direct dependency. Establish what is known about the relevant runtime or build use without treating mere installation as proof of exploitability. The next action can be a parent-package upgrade, an application compatibility check or a manual vendor review.
A missing or failed report is an unknown result, not a clean bill of health.
Keep automated fixes separate
npm audit fix can modify the dependency tree and invokes installation behavior. It is a separate change from reporting. The --audit-level option controls the exit threshold rather than hiding lower-severity report entries.
Agree the CI policy explicitly and evaluate changes in a review branch before adopting them. For a planned reproducible installation, npm ci requires lockfile agreement and does not rewrite the lockfile; rebuilding consistently is still distinct from proving a trusted environment.
The npm-scanner finding needs a different route
OSV published MAL-2026-19071 for the exact npm-scanner package on 10 November 2026. Its record identifies malicious code with no fixed version listed. It is not a finding against npm CLI as a whole.
The source recommends treating a machine with that package installed or running as compromised and recovering credentials from another machine. Package removal alone is not evidence that the environment is trustworthy again.
Agree the incident boundary before a rebuild
If exact-package installation is confirmed, ask the authorized incident or CI owner to establish the affected runner population, accessible resources and artifact lineage. Keep that work separate from an ordinary audit-fix change. Agree a trusted replacement environment, reviewed dependency inputs and which generated client assets need revalidation.
Do not upload secrets or full build logs to a public support form. The article does not claim that any particular WordPress project used this package.
Close with evidence the client can understand
Use a small review record with one row for the dependency finding, one for the agreed change or recovery scope, and one for the client journey or build acceptance. Include unresolved questions and the owner of each next decision. Compare report evidence for the same intended build rather than comparing unrelated workstations.
Skepra published this original agency handoff framework on 11 November 2026; no customer environment was scanned or recovered for the article.