devez-vibe@1.9.52
Stable terminal UI for Codex and Claude Agent SDK
AI Security Review
scanned 10d ago · by lpm-firewall-aiThe inspected source confirms automatic installation-time setup of package-owned agent skills, with replacement or pruning possible within those skill directories. That warrants a warning for agent extension lifecycle risk, but the inspected behavior does not establish a broader control-surface hijack or other concrete attack.
Decision evidence
public snapshot- The npm postinstall hook runs both the skill installer and the dependency preparation script.
- The installer targets the user's Codex and Claude skill directories and includes the luna-loop and insane-search skills.
- For each target, the installer copies the bundled skill, removes destination items absent from the bundle, and writes a bundled marker.
- The bridge's permission configuration can write to Claude user, project, or local settings files.
- Permission changes are exposed through a runtime permissions/update request; this is separate from the npm postinstall hook.
- The automatic agent-directory setup is limited to two package-owned skills; the inspected installer does not write broad agent settings or control files.
- The bridge's permission and plugin operations are runtime request handlers, not actions invoked by the postinstall script.
Source & flagged code
8 flagged · loading sourcePackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgSource creates an unconsented AI-agent control surface through install-time mutation or a default unauthenticated remote skill channel.
Findings
Affected versions and remediation
This report applies to devez-vibe@1.9.52.
See all versions of devez-vibe for other recorded verdicts.
- Review the evidence and your use of devez-vibe@1.9.52 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.