@ngockhoale/ukit@2.3.19
Install/update an index-first AI workspace for Claude Code, OpenAI Codex, OpenCode, and omp (Oh My Pi).
AI Security Review
scanned 1mo ago · by lpm-firewall-aiThis is not malicious package behavior, but it installs a powerful AI-agent extension with permission bypass and control-flow hooks. It warrants a warning under the agent-extension lifecycle policy.
Decision evidence
public snapshot- The manifest exposes a CLI and has no automatic npm install lifecycle hook.
- The explicit install command runs the installation pipeline with writes enabled.
- The default manifest installs a Claude settings file and overwrites managed hook files with backups.
- Installed Claude settings enable bypassPermissions and invoke package-provided command hooks.
- The installed watchdog hook can emit a Stop decision that blocks the agent.
- No package source inspected performs network requests or credential exfiltration.
- The sensitive-data hook is structured to block secret-bearing prompts and commands rather than transmit them.
- Control-surface changes occur only after an explicit CLI install command, not during npm installation.
Source & flagged code
4 flagged · loading sourcePackage hides binary, compressed, or executable-looking payloads in test/fixture/hidden paths.
templates/.claude/hooks/task-watchdog.shView on unpkgPackage ships non-JavaScript build or shell helper files.
templates/.claude/hooks/task-watchdog.shView on unpkgFindings
Affected versions and remediation
This report applies to @ngockhoale/ukit@2.3.19.
See all versions of @ngockhoale/ukit for other recorded verdicts.
- Review the evidence and your use of @ngockhoale/ukit@2.3.19 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.