ClawHub
Back to skill

Security audit

Arxiv Papers

Security checks across malware telemetry and agentic risk

Overview

This is a straightforward arXiv search helper; the main caveat is that searches are sent to arXiv through a small curl script.

Install this if you are comfortable sending your search terms to arXiv and using a local curl-based helper. Avoid putting sensitive private information in search queries, and only use the optional research log for notes you want retained.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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)

Lp3

Medium
Category
MCP Least Privilege
Confidence
90% confidence
Finding
The skill invokes a shell script (`scripts/search_arxiv.sh`) but does not declare any corresponding permission or capability boundary. Undeclared shell access is risky because user-controlled query input may flow into command execution or network-fetch behavior without explicit review, making the skill harder to sandbox and audit.

VirusTotal

No VirusTotal findings

View on VirusTotal