ClawHub
Back to skill

Security audit

Geo Ai Plugin Builder

Security checks across malware telemetry and agentic risk

Overview

This skill is a planning/template assistant for GEO-aware AI plugin design, with no evidence of hidden installs, credential access, persistence, or destructive behavior.

Safe to install as a design assistant. Review generated plugin specs before building them, especially for authentication, privacy, retention, regulated data, telemetry, and overly broad tool invocation rules.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Vague Triggers

Medium
Confidence
94% confidence
Finding
The manifest description uses expansive trigger language such as 'whenever the user mentions building, designing, packaging, or standardizing AI plugins/tools,' which can cause the orchestrator to activate on loosely related requests. Over-broad auto-invocation increases the chance this skill overrides more appropriate domain-specific skills, causing scope hijacking, irrelevant guidance, or unsafe handling of specialized contexts like regulated or sensitive plugin designs.

Natural-Language Policy Violations

High
Confidence
98% confidence
Finding
The phrase 'users want to完成' introduces forced non-English output or mixed-language behavior without user consent. While likely accidental, this can degrade reliability, confuse users, and cause malformed downstream artifacts when consumers expect consistent English-only specs or copy-paste-ready technical text.

Vague Triggers

Medium
Confidence
90% confidence
Finding
The guidance recommends descriptions like 'Use this tool whenever the user asks to compare…' and 'Call this tool when the user provides concrete numbers and wants…' without emphasizing narrower eligibility checks, exclusions, or confidence thresholds. In a master orchestrator for plugin/tool building, this can lead downstream tools to be invoked too broadly, causing unintended data exposure, irrelevant actions, or overuse of tools in ambiguous situations.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.