Security audit

CouponClaw

Security checks across malware telemetry and agentic risk

Overview

CouponClaw is a disclosed coupon and deals assistant with optional daily deal notifications, not evidence of malware or hidden data theft.

Install only if you want an assistant that can browse public coupon/deal sites and, if you explicitly enable it, create recurring daily deal notifications. Review and use the off/status commands for any push subscription you enable.

SkillSpector

By NVIDIA

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Tp4

High

Category: MCP Tool Poisoning
Confidence: 93% confidence
Finding: The skill’s declared purpose centers on coupon discovery and final-price calculation, but the file also defines daily-deals generation and scheduled push delivery to external messaging platforms. That is a meaningful capability expansion because it introduces automation, outbound messaging, and persistent scheduling behavior that users may not reasonably infer from the headline description, increasing the risk of unexpected actions or spam-like behavior.

Description-Behavior Mismatch

Medium

Confidence: 94% confidence
Finding: This script implements persistent push subscription and cron scheduling, which is a material capability beyond the stated coupon-search and price-calculation behavior in the skill description. Even though the code validates inputs and does not show command injection, undisclosed scheduled messaging expands the skill’s reach and could surprise users, create unsolicited notifications, or be repurposed for ongoing outbound contact.

Vague Triggers

Medium

Confidence: 90% confidence
Finding: The invocation guidance includes broad shopping phrases such as requests about saving money or best final price, which can overlap with normal commerce conversations. In an agent environment, this can cause unintended invocation and let the skill browse coupon/deal sites or steer the interaction when the user did not actually ask for coupon hunting.

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The trigger list contains highly generic phrases like 'daily deals', 'best deals today', and broad multilingual terms for discounts or saving money. These are insufficiently constrained and could activate the skill in many unrelated shopping contexts, especially since the skill also supports deal briefings and push features that extend beyond simple coupon lookup.

Vague Triggers

High

Confidence: 96% confidence
Finding: The trigger list includes broad, everyday shopping phrases such as 'coupon', 'save money', 'daily deals', and common Chinese equivalents, which can cause the skill to activate on many ordinary user requests not specifically intended for this package. Over-broad activation increases the chance of unwanted invocation, context hijacking, and accidental exposure of users to this skill's behavior when they were seeking general advice rather than this particular tool.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.