@ngockhoale/ukit@2.2.5
Install/update an index-first AI workspace for Claude Code, OpenAI Codex, OpenCode, and omp (Oh My Pi).
AI Security Review
scanned 2mo ago · by lpm-firewall-aiNo malicious chain was found, but explicit installation deploys an AI-agent control extension with bypass-permissions settings and executable hooks. This is a guarded first-party extension risk, not an npm lifecycle compromise.
Decision evidence
public snapshot- The explicit CLI install path writes a Claude Code settings file and many executable hook templates into the current project.
- The installed Claude settings select bypass-permissions mode and invoke hooks for edit, shell, and prompt events.
- A prompt hook reads prompt data and starts a local Node helper; it does not itself download URL images.
- This is a first-party AI workspace extension installed only when the user runs the CLI, not during npm installation.
- package.json has no preinstall, install, or postinstall lifecycle hook.
- No network client or outbound request implementation was found in executable package source.
- The reviewed shell hook blocks destructive shell patterns rather than executing them.
- The installer writes through a manifest and tracks managed project paths; no credential collection or exfiltration was found.
Source & flagged code
5 flagged · loading sourcePackage hides binary, compressed, or executable-looking payloads in test/fixture/hidden paths.
templates/.claude/hooks/vision-router.shView on unpkgPackage ships non-JavaScript build or shell helper files.
templates/.claude/hooks/vision-router.shView on unpkgFindings
Affected versions and remediation
This report applies to @ngockhoale/ukit@2.2.5.
See all versions of @ngockhoale/ukit for other recorded verdicts.
- Review the evidence and your use of @ngockhoale/ukit@2.2.5 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.