ClawHub
Back to skill

Security audit

Agentic Commerce News

Security checks across malware telemetry and agentic risk

Overview

This is mostly a public news-briefing skill, but it also tells agents to create recurring cron-style jobs using tools not declared in its manifest.

Use it for one-off public news briefings with normal web access. Before enabling any daily or recurring digest, confirm exactly which scheduler will be used, avoid system crontab unless you intend local persistence, and make sure you know how to list and remove the scheduled job.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (4)

Context-Inappropriate Capability

Medium
Confidence
92% confidence
Finding
The skill is declared as a web-only news briefing tool, but its instructions attempt to expand behavior into scheduling and system-cron creation. That creates capability confusion and can cause an agent/runtime with broader ambient permissions to perform persistence or command-execution actions the user did not expect from this skill.

Intent-Code Divergence

Medium
Confidence
95% confidence
Finding
The documented behavior promises scheduling and persistence flows that are not reflected in the manifest's declared tools, creating a mismatch between claimed scope and actual behavior. This kind of contradiction is dangerous because it weakens policy enforcement, misleads reviewers, and may encourage an orchestrator to satisfy the instructions through unintended tools or ambient system access.

Vague Triggers

Medium
Confidence
91% confidence
Finding
The trigger conditions are broad enough to activate on ordinary requests about news, updates, or scheduling, which can cause the skill to run without a clear explicit user request for this specific capability. In a tool-using skill, overbroad auto-triggering increases the chance of unintended web access, scheduled task creation, or surprising behavior that the user did not specifically authorize.

Vague Triggers

Medium
Confidence
83% confidence
Finding
The activation text includes broad trigger phrases such as general requests for news or daily digests, which can cause the skill to auto-invoke in situations where the user did not intentionally select it. Because the skill also contains scheduling instructions, over-broad activation increases the chance of unintended recurring actions or unnecessary web access.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.