Llava Vision

Security checks across malware telemetry and agentic risk

Overview

This skill does what it claims: it sends a user-selected image to a local LLaVA llama.cpp server for analysis.

Install this only if you intend to use and trust a local llama.cpp/LLaVA server. Be deliberate about image paths and URLs: selected local images and prompts are sent to the local server, and remote image URLs will be fetched before analysis.

SkillSpector

By NVIDIA

Vulnerability Patterns

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (1)

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill sends user-supplied image content to an HTTP API endpoint without any disclosure, consent mechanism, or transport security. Even though the destination is localhost, this still transfers potentially sensitive local file contents or remote-fetched images to another service process, which can expose private data to unintended logging, interception by local malware, or misuse by a differently configured local server.

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal