@apilift/cli@1.4.39
AI Security Review
scanned 20d ago · by lpm-firewall-aiSource inspection resolved the scanner hints to schema validation, fixed local process inspection, user-invoked agent setup, and usage measurement collection. Credential setup is explicit and local, and app requests are constrained to the selected transport's allowed origins.
Decision evidence
public snapshot- The manifest exposes CLI binaries and only build, typecheck, and test scripts; it has no install lifecycle hook.
- Setup submits a one-time code supplied by the Apilift setup page to the configured registry and saves the returned installation token. App credentials use an explicit secure setup flow and are scoped to declared app destinations.
- Agent plugin installation is reached through the explicit setup-continue command and uses local agent plugin commands; the manifest has no install hook that mutates agent configuration.
- The indicated session-audit location is tokenizer data. The audit command requires a supplied transcript or session and writes a local report.
- The collector reads transcript paths to reconcile usage facts, then sends derived measurements to the Apilift service; the payload contains usage metadata rather than transcript text.
- The flagged dynamic evaluation generates schema validators, the capture worker uses fixed process-inspection code, and the native host checks its caller against the configured extension origin.
- The scanner's remote-code-execution hint is not supported by the inspected capture worker: its evaluated source is a fixed process metadata script.
Source & flagged code
5 flaggedFindings
Affected versions and remediation
This report applies to @apilift/cli@1.4.39.
See all versions of @apilift/cli for other recorded verdicts.
- Review the evidence and your use of @apilift/cli@1.4.39 before allowing it. Restrict the permissions described in this report.
- Choose an independently verified alternative or release. This report does not establish that other versions are safe.
Evidence last updated: .
Think this verdict is wrong? Dispute it on GitHub or read how reports are produced.