Security audit

ZM IMG2 视觉提示词编排

Security checks across malware telemetry and agentic risk

Overview

This skill appears to be a user-directed image-prompt compiler with optional image-generation execution, but users should verify the downstream image skill it invokes.

Before installing, confirm that the expected downstream image skill is installed and reviewed, especially because the metadata and script use different dependency names. Treat any reference images and prompts as data that may be sent to the configured Happy/OpenAI-compatible image provider when you run with --run.

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: 88% confidence
Finding: The skill declares no permissions, yet its documented behavior requires reading local files, writing artifacts under the OpenClaw home directory, and invoking Python scripts via the shell. This creates a trust and review gap: operators may authorize or run the skill without understanding that it can access the filesystem and execute commands, which increases the chance of unintended file exposure or unsafe command execution in the broader agent environment.

VirusTotal

64/64 vendors flagged this skill as clean.

View on VirusTotal