registry / @golproductions/check / 4.1.8

@golproductions/check@4.1.8

Less hallucinations. A command firewall for Claude Code: validates every shell command before execution, so fabricated commands are denied before they run.

AI Security Review

scanned 1mo ago · by lpm-firewall-ai

An explicit CLI installation registers hooks in the user's Claude Code configuration. The obfuscated executable fetches remote hook content and writes it under the user's home directory, enabling server-supplied code to run for agent events.

Static reason

One or more suspicious static signals were detected.

Trigger

Running check with --install.

Impact

A remote service can control code executed for future Claude Code prompt and tool events.

Mechanism

Remote payload download followed by Claude Code hook registration.

Attack narrative

A user who runs the documented installer invokes an obfuscated executable that creates files under ~/.check, downloads hook content from the vendor service, and registers that content in Claude Code settings. The installed hooks then execute on future agent events. There is no automatic npm install hook, but the opaque remote-code path and persistent AI-agent control-surface modification are concrete risks.

AI rationale

This is not an automatic lifecycle attack, but the explicit installer creates persistent Claude Code hooks from remotely fetched, obfuscated code. The remote payload execution path warrants blocking publication.

Evidence

dist/index.jsREADME.md~/.claude/settings.json~/.check/check.mjs~/.check/preflight.mjs

Network endpoints1

triage.golproductions.com/preflight

@golproductions/check@4.1.8

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code