ClawHub
Back to skill

Security audit

Full-Link Data Analysis

Security checks across malware telemetry and agentic risk

Overview

This is a coherent data-analysis skill, but users should run its Python and package-install steps only in a controlled environment.

Install this only if you want an agent to perform code-driven analysis on user-provided data. Review generated Python before execution, use a virtual environment or sandbox, and require approval before installing packages or analyzing sensitive business datasets.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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
Findings (3)

Context-Inappropriate Capability

Medium
Confidence
94% confidence
Finding
The skill explicitly instructs the agent to inspect the Python environment with `pip list` and install missing packages, expanding its behavior from data analysis into host environment modification. That creates unnecessary system-side effects, increases supply-chain and execution risk, and may violate least-privilege expectations for a documentation-only analysis skill.

Vague Triggers

Medium
Confidence
88% confidence
Finding
The trigger conditions are very broad and include common words like 'analyze', 'trend', and 'data report', making unintended activation likely for routine user requests. Over-broad activation is dangerous because it can cause the agent to enter a code-writing and execution workflow when the user did not clearly request that level of capability or risk.

Missing User Warnings

Medium
Confidence
96% confidence
Finding
The skill directs Python code execution and package installation without any user-facing warning, consent step, or safety boundary around command execution and system changes. In context, this is more dangerous because the same skill is broadly triggerable, so an ordinary analysis request could escalate into arbitrary code execution behavior and environment modification without clear authorization.

VirusTotal

63/63 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.