ClawHub
Back to skill

Security audit

DynaPictures

Security checks across malware telemetry and agentic risk

Overview

The skill is a coherent DynaPictures connector, but it under-describes image generation as a safe/read-style action even though it can create hosted outputs and use credits.

Review before installing. This skill should require explicit approval before `generate_images`, including the template, payload, expected credit use, and where hosted image URLs may be exposed. Listing or retrieving templates is lower risk.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (3)

Description-Behavior Mismatch

Medium
Confidence
95% confidence
Finding
The manifest says the skill is for 'searching and reading data,' but the documented actions include `generate_images`, which performs content creation rather than a read-only operation. This mismatch can mislead an agent or user into treating the skill as low-risk and invoking it without the scrutiny or confirmation appropriate for a non-read action.

Intent-Code Divergence

Medium
Confidence
98% confidence
Finding
The safety section states that untagged actions are reads and safe to run directly, yet `generate_images` is untagged despite being a non-read operation that can consume credits and create hosted outputs. This unsafe classification can cause automatic execution without user confirmation, leading to unintended spending or generation of sensitive or policy-violating content.

Vague Triggers

Medium
Confidence
87% confidence
Finding
The manifest uses an overly broad trigger phrase: 'Use this skill for ANY DynaPictures request.' Broad routing language increases the chance an agent will invoke this skill in situations where direct API use, narrower tooling, or additional validation would be more appropriate. In this file, that risk is amplified because the skill is also inaccurately presented as read-oriented while exposing a generation action.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.