ClawHub
Back to skill

Security audit

Data Analyst Expert Free

Security checks across malware telemetry and agentic risk

Overview

This skill is a straightforward data-analysis helper that uses local files and code execution in ways that match its stated purpose.

Install only for data files you are comfortable processing in the agent workspace. Review generated analysis code before running it on untrusted datasets, keep outputs in the working directory, and provide external API credentials only through environment variables if you intentionally add such integrations.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • 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
Findings (2)

Intent-Code Divergence

Low
Confidence
71% confidence
Finding
The skill strongly states that file access should be limited to filenames in the working directory, but later expands capabilities to create/query/export operations without restating equivalent path constraints. In a skill with write and exec permissions, ambiguous export/create behavior can lead to unsafe file writes, path traversal, or accidental overwrite of unintended files if downstream agents infer broader filesystem access.

Missing User Warnings

Medium
Confidence
88% confidence
Finding
The skill advertises file writes and arbitrary code execution as core capabilities without an upfront warning about risks such as executing untrusted code, overwriting files, exposing sensitive data, or producing side effects in the runtime. In an agent setting, this can normalize dangerous delegation and increase the chance that users or orchestrators invoke the skill on untrusted inputs without adequate safeguards.

VirusTotal

61/61 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.