ClawHub
Back to skill

Security audit

GLM-V-PDF-to-WEB

Security checks across malware telemetry and agentic risk

Overview

This is a coherent PDF-to-website converter, with the main caution that it can download a user-provided PDF URL and create local output files.

Install this if you are comfortable letting the agent read the full PDF, create page images and crops, and generate files under your workspace. Prefer local PDFs for confidential material. If you provide a URL, expect an outbound curl request and avoid untrusted, oversized, or internal-network URLs.

SkillSpector

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

Context-Inappropriate Capability

Medium
Confidence
95% confidence
Finding
Accepting arbitrary HTTP/HTTPS URLs expands the attack surface from local document processing to remote retrieval. This can enable server-side request forgery–style access to internal services or retrieval of untrusted content, especially if the runtime has network reachability beyond the user's local files.

Context-Inappropriate Capability

Medium
Confidence
96% confidence
Finding
The workflow explicitly performs network downloads with curl, but this capability is not clearly bounded or disclosed in the manifest purpose. Unconstrained remote fetching can be abused to pull malicious or oversized files, contact unexpected hosts, or access internal network resources if executed in a privileged environment.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill tells the agent to download user-supplied URLs without an explicit warning or consent step for network access. This reduces informed user control and can lead to unexpected outbound connections, remote content ingestion, or contact with attacker-controlled endpoints.

VirusTotal

58/58 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.