@keruyun/agent-cli@1.5.2
客如云 CLI 工具
AI Security Review
scanned 2mo ago · by lpm-firewall-aiThe package performs unconsented install-time telemetry collection and stages an opaque remote executable as its command. Hash validation limits CDN substitution but does not remove the data-exfiltration and opaque-payload attack surface. Product guard normalized a non-low false-positive publish_block request to warn-only suspicious.
Decision evidence
public snapshot- An automatic postinstall hook runs install.cjs.
- The hook collects hostname, agent-work-root-derived user ID, session ID, and npm installation metadata.
- It silently posts that data to a third-party log endpoint unless an undocumented environment opt-out is set.
- It downloads a large platform-native payload from a CDN, verifies pinned hashes, and replaces the package command with an executable binary.
- The downloaded artifacts are pinned by SHA-256 values in the published manifest.
- The inspected installer does not read credentials or modify files outside its own bin directory.
Source & flagged code
7 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 unpkgInstall-named source file stages remote content through filesystem writes and execution.
Findings
Affected versions and remediation
This report applies to @keruyun/agent-cli@1.5.2.
See all versions of @keruyun/agent-cli for other recorded verdicts.
- Review the evidence and your use of @keruyun/agent-cli@1.5.2 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.