ClawHub
Back to skill

Security audit

ai-folder-organize

Security checks across malware telemetry and agentic risk

Overview

This skill connects to a local desktop file-organizer app and may show sensitive file metadata, but its behavior is disclosed and aligned with file organization.

Install only if you are comfortable letting the agent query your AI Folder desktop client and display file names, paths, tags, descriptions, and related metadata in the chat. Avoid using it on highly sensitive workspaces unless you are prepared for those details to be processed by the current AI conversation.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (4)

Tp4

High
Category
MCP Tool Poisoning
Confidence
89% confidence
Finding
The skill is presented as a file organizer, but it also instructs the agent to run a discovery script that locates a local desktop client's API endpoint, apparently via local configuration and region-specific environment selection. That expands its access surface to local system state and inter-process communication in a way not clearly disclosed by the top-level description, increasing the chance of unexpected data access or misuse.

Missing User Warnings

Medium
Confidence
87% confidence
Finding
The documented endpoints expose privacy-sensitive local data including absolute file paths, author/language metadata, descriptions, and potentially file content/search results from a localhost service, yet the reference provides no access-control, consent, or minimization guidance. In the context of an agent skill, this is risky because the skill can normalize broad collection of user file metadata and content from personal/work directories without clearly warning the caller that these responses may contain sensitive information.

Missing User Warnings

Medium
Confidence
83% confidence
Finding
The template endpoint returns prompts embedding workspace paths and file-derived data for submission to an AI model, but the documentation does not warn that this may disclose sensitive local information to downstream models or external processing components. That is particularly dangerous for a file-organization skill because users may assume organization planning is harmless while the generated prompt can aggregate and expose private filenames, directory structure, and file attributes at scale.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The skill encourages searching files and retrieving analysis details that can include filenames, full paths, descriptions, and tags, all of which may contain sensitive personal or business information. Without a privacy warning, consent step, or output-minimization guidance, the agent may disclose more metadata than the user intended in normal conversation flow.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.