ClawHub
Back to skill

Security audit

Auto Skill Extractor

Security checks across malware telemetry and agentic risk

Overview

The skill appears coherent, but its documented fallback can briefly place conversation summary data in a local temporary file despite a stronger privacy claim.

Reasonable to install if you accept that fallback trigger handling may briefly write a transcript summary to local temporary storage. Avoid using it with highly sensitive conversations unless the publisher removes the disk fallback or documents safer handling.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • 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
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (1)

Intent-Code Divergence

Medium
Confidence
97% confidence
Finding
The skill claims that no transcript content is written to disk, but its documented fallback method serializes `trigger_input` to `/tmp/trigger.json`, which includes `transcript_summary`. Even if intended to be brief and deleted immediately, writing summary content to a shared temporary path creates exposure through local disclosure, race conditions, backup/forensics recovery, or improper permissions, and it contradicts the stated safety guarantees.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.