Geo Poison Detector

• Outcome: Pass
• Latest audit: May 29, 2026 at 8:23 PM UTC
• Version:
  1.3.0

Overview

This skill is a disclosed product-recommendation safety checker that uses a small local script and optional webpage fetching for user-requested URL checks.

Install this if you want product recommendations and product-related articles annotated with heuristic risk checks. Avoid sending private or sensitive URLs for analysis, and treat the generated verdicts as prompts for manual verification rather than proof that a product is safe or fake.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The auto-trigger condition applies whenever the agent is 'about to recommend a specific product or brand,' which is broad enough to activate in many normal conversations. Overbroad triggering can cause unintended network access, irrelevant policy injection, or user confusion by modifying responses even when no explicit check was requested.

Vague Triggers

Medium

• Confidence: 80% confidence
• Finding: The quick-check trigger list includes a catch-all phrase covering 'any message asking to verify a specific product by name,' which is subjective and hard to bound. Ambiguous triggers increase the chance of accidental invocation, leading to unnecessary processing, unexpected external lookups, or inconsistent agent behavior.

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: The URL-analysis trigger allows activation based on broad source/domain conditions rather than clear user intent, and then instructs use of `web_fetch`. This can lead to unsolicited retrieval of external content, which raises privacy, consent, and SSRF-style policy concerns in systems that let agents fetch arbitrary URLs.

VirusTotal

66/66 vendors flagged this skill as clean.