@jive-sh/skill-sync@0.1.2
Keeps one skills directory identical across every machine you work on.
AI Security Review
scanned 14d ago · by lpm-firewall-aiThe postinstall automatically registers a persistent per-user daemon, creating residual lifecycle setup risk despite its CI and repair-only guards. The inspected source supports a warning for package-owned extension lifecycle behavior, not a confirmed malicious attack.
Decision evidence
public snapshot- package.json runs install.js during postinstall and declares platform-specific optional packages under aliases of this package.
- The installer repairs command targets, registers daemons, and removes its install script during postinstall.
- On macOS, the installer creates a login agent that starts at login and restarts after unsuccessful exits.
- On Linux, the installer enables and starts a per-user systemd unit immediately.
- The installer skips daemon registration only for repair-only runs or when CI is set; otherwise it registers the package daemon.
- The package detects Claude skills directories, and its CLI syncs files using a saved token and configured endpoint.
- The daemon registration is scoped to the current user and package-owned daemon; the installer includes CI and repair-only skip conditions.
- Inspected source shows no unrelated recipient or credential transfer; the sync caller sends its saved token to the configured endpoint.
Source & flagged code
6 flaggedPackage defines install-time lifecycle scripts.
package.jsonView on unpkgInstall-time lifecycle script is not statically allowlisted and needs review.
package.jsonView on unpkgFindings
Affected versions and remediation
This report applies to @jive-sh/skill-sync@0.1.2.
See all versions of @jive-sh/skill-sync for other recorded verdicts.
- Review the evidence and your use of @jive-sh/skill-sync@0.1.2 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.