Clash of Coins - Agentic Gateway

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed Clash of Coins checkout guide with high-impact payment steps, but the artifacts do not show hidden execution, credential theft, persistence, or automatic purchases.

Install this only if you intend to let an agent help with Clash of Coins discovery and checkout routing. Treat any quote, buy, order creation, wallet funding, or payment retry as a real spending action; verify the origin, item, recipient, amount, protocol, and endpoint yourself before allowing the agent to proceed.

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 (2)

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill provides concrete purchase and payment execution steps, including buy endpoints and agent-wallet funding flows, but does not require an explicit user confirmation or warn that these actions can spend funds and create real purchases. In an agent context, this increases the risk of unintended financial transactions if the model interprets browsing or integration tasks as authorization to execute payment-capable flows.

Vague Triggers

Medium
Confidence
96% confidence
Finding
The trigger example explicitly asks the skill to route a purchase when the user 'did not say sale or shop', which undermines the skill's stated safety boundary of not mixing sale and shop contracts. In a purchasing/integration skill, ambiguous invocation can cause the agent to infer intent incorrectly and drive users into the wrong contract flow, increasing the risk of unintended purchases or protocol misuse.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal