ClawHub
Back to skill

Security audit

Save Money

Security checks across malware telemetry and agentic risk

Overview

This is a transparent cost-routing skill that may automatically forward complex prompts to a stronger Claude session, but it does not show hidden execution, persistence, credential access, or malicious behavior.

Install this only if you want automatic model handoff for complex or uncertain prompts. Avoid using it with sensitive prompts unless you are comfortable with the full task being sent to a spawned Sonnet session, and watch costs because escalation does not ask for separate confirmation.

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)

Vague Triggers

Medium
Confidence
95% confidence
Finding
The escalation examples are intentionally broad and include common requests such as analysis, planning, comparison, summarization, and writing. In a behavioral skill that is injected into model instructions, this can systematically trigger `sessions_spawn` for ordinary user prompts, causing unnecessary full-task forwarding to a stronger model and expanding data exposure beyond what the user explicitly requested.

Vague Triggers

Medium
Confidence
92% confidence
Finding
The 'more than 30 seconds of focused thinking' rule is subjective and impossible to apply consistently, which encourages over-escalation. Because the skill says to escalate when in doubt and to do so immediately with the full task, this ambiguity can be exploited to route a wide range of benign prompts to another model without clear necessity.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.