ClawHub
Back to skill

Security audit

Content Parser

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed URL-content extractor that uses an external API and can save results locally, with no evidence of hidden or destructive behavior.

Install only if you trust the ListenHub/Marswave API with the URLs you submit and are comfortable using a LISTENHUB_API_KEY. Disable autoDownload or avoid this skill for private, internal, signed, regulated, or confidential links if you do not want extracted content and raw API responses saved in the working directory.

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)

Description-Behavior Mismatch

Medium
Confidence
87% confidence
Finding
The skill’s stated purpose is URL content extraction, but it also creates persistent local state and saves extracted artifacts by default. That expands the data-handling surface beyond what a user would reasonably expect, increasing the risk that sensitive extracted content or metadata is left on disk without informed consent.

Context-Inappropriate Capability

Medium
Confidence
85% confidence
Finding
The skill persists both configuration and extracted artifacts locally even though its core purpose is remote URL parsing. Persistent storage increases exposure of potentially sensitive content, creates residual data on the host, and is only loosely justified in the current documentation.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The skill auto-saves full extracted content and raw API responses to disk without a prominent user-facing warning that sensitive source material, metadata, or tokens embedded in responses may be retained locally. This is especially risky for private or regulated content because the saved JSON may contain more information than the user expects.

Vague Triggers

Medium
Confidence
95% confidence
Finding
The `General web | Any HTTP(S) URL` pattern is overly broad for a skill whose trigger is parsing user-provided URLs, because it can match almost any web link and may cause the skill to activate in situations beyond the user's clear intent. In an agent environment, this increases the risk of over-triggering, unintended network access, and processing of untrusted external content that other skills or conversations mention incidentally.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.