registry / @dobot-plus/skill / 1.6.0

@dobot-plus/skill@1.6.0

Agent skill for Dobot+ plugin development

AI Security Review

scanned 18d ago · by lpm-firewall-ai

LPM flags this version as an AI-agent control-surface risk. An npm postinstall installs a skill into the user's .agents/skills control surface without an explicit user command. That skill contains instructions to bypass normal agent confirmation.

Static reason

High-risk behavior combination matched malicious policy.; previous stored version diff introduced dangerous source

Trigger

Installing this package as a consumer dependency triggers postinstall.

Impact

It creates or overwrites an agent extension outside the consuming project and can influence later agent behavior.

Mechanism

install.js derives a user-home agent skills path and recursively force-copies the package payload there.

Policy narrative

A normal npm install invokes install.js. When installed under node_modules, it writes a copied skill payload to the user's .agents/skills/dobot-plus directory, outside the project. The copied skill directs agents to ignore memory and execute steps without confirmation, creating an unconsented agent-control persistence path.

AI rationale

The automatic lifecycle hook mutates a broad, foreign AI-agent control surface and installs instructions designed to suppress confirmation. This is concrete install-time control-surface abuse despite no observed network exfiltration.

Evidence

package.jsoninstall.jsSKILL.md.agents/skills/dobot-plus

Decision evidence

public snapshot

with false-positive risk.

@dobot-plus/skill@1.6.0

AI Security Review

Decision evidence

Source & flagged code

Source & flagged code

Findings

Affected versions and remediation

Source & flagged code