Security audit

Blender Bpy Enhanced

Security checks across malware telemetry and agentic risk

Overview

This is a coherent Blender automation skill, with expected scene-generation and render/export behavior, but users should run the demo only in a new or disposable Blender scene.

Install only if you want Blender automation help. Treat the bundled demo and clear-scene snippet as destructive to the current Blender scene: run them in a new, saved, or disposable .blend file, and choose explicit output filenames when rendering or exporting to avoid overwrites.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (7)

Description-Behavior Mismatch

Medium

Confidence: 95% confidence
Finding: The script immediately deletes all objects and removes all materials from the current Blender scene, which is destructive behavior unrelated to a narrowly scoped gear-generation action. It also writes output files to fixed /tmp paths later in the script, so running it can cause data loss in the active scene and undisclosed filesystem side effects without user consent.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The documented scene setup unconditionally selects and deletes all objects, then removes all materials from the current Blender file. In a Blender automation skill, this is plausible functionality, but without an explicit warning or safer scoping it can cause unintended destructive loss of user work when run in the wrong project or context.

Missing User Warnings

Low

Confidence: 87% confidence
Finding: The render helper writes an image file to a fixed path by default and will overwrite or create artifacts on disk without any user-facing warning. This is low severity, but in automated environments it can unexpectedly modify the filesystem or clobber existing outputs if the path is reused.

Missing User Warnings

Low

Confidence: 86% confidence
Finding: The STL export example creates output artifacts on disk and does so without warning about filesystem side effects. In context this is expected behavior for an export skill, but users should still be informed because it can overwrite files or leave artifacts in shared workspaces.

Missing User Warnings

Medium

Confidence: 96% confidence
Finding: The script immediately selects and deletes all objects and removes all materials from the current Blender scene without warning, confirmation, or scope limitation. In the context of a reusable automation skill, this is dangerous because running the demo against an active user project can cause irreversible loss of work or corruption of the working scene state.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The script performs unconditional scene-wide deletion at startup via select-all and delete, then removes all materials, with no prompt or safeguard. In a Blender automation skill, this is dangerous because users may execute it in an existing project and lose unsaved or valuable scene state unexpectedly.

Missing User Warnings

Low

Confidence: 90% confidence
Finding: The script saves a .blend file and configures rendering to write an image to disk without prior disclosure or user approval. Although the target is /tmp and impact is lower than arbitrary-path writes, it still creates filesystem side effects and could overwrite prior temporary files or surprise users in restricted environments.

VirusTotal

62/62 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.