registry / @lovrabet/rabetbase-cli / 2.5.4

@lovrabet/rabetbase-cli@2.5.4

Developer CLI for Lovrabet apps, datasets, pages, SQL, Backend Function, database connections, skills, and AI agent workflows.

AI Security Review

scanned 9d ago · by lpm-firewall-ai

Installation automatically launches an unpinned npx skill installer with global scope. This can modify shared AI-agent skill locations in the user's home directory.

Static reason

High-risk behavior combination matched malicious policy.

Trigger

npm postinstall during package installation.

Impact

A package installation can alter broad, shared AI-agent control surfaces and fetch current external installer code without an explicit user command.

Mechanism

The lifecycle hook runs the built-in skill installer, which calls npx with skills@latest, an add command, and a global flag.

Attack narrative

The package uses postinstall to invoke an obfuscated built-in installer. That installer automatically runs an unpinned npx skills@latest command with global scope to install a Lovrabet skill into shared agent-skill locations. The action occurs without an explicit setup command and can modify multiple AI-agent environments in the user's home directory.

AI rationale

The automatic lifecycle chain mutates a broad AI-agent control surface and downloads current external installer code through npx. This meets the install-hook abuse blocking boundary.

Evidence

package.jsonscripts/postinstall.mjslib/postinstall.jslib/skills/builtin-skill.jslib/skills/npx-skills-add.jslib/skills/skill-presence.js.agents/skills.claude/skills

@lovrabet/rabetbase-cli@2.5.4

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code