ClawHub
Back to skill

Security audit

Comfyui Local

Security checks for vulnerabilities and agentic risk

Overview

This skill appears to be a purpose-aligned local ComfyUI image generator, with some disclosed persistence and resource-use considerations but no evidence of deception, exfiltration, or destructive behavior.

Install only if you intend to send prompts to a local or private ComfyUI server you trust. Expect generated images to be written to disk, and review or clean both the skill output folder and ComfyUI output folder if prompts or images are sensitive. The bundled workflows may require specific ComfyUI custom nodes and models and can use significant GPU/CPU resources.

SkillSpector

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • 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

High
Confidence
94% confidence
Finding
The allowlisted workflow contains substantial video upscaling functionality that is not reflected in the stated skill purpose of local image generation. Hidden or undocumented processing capabilities expand the attack surface and can enable resource abuse, unexpected handling of user-supplied media, and execution of code paths that operators did not intend to expose.

Context-Inappropriate Capability

High
Confidence
95% confidence
Finding
Including video-processing components in an image-generation skill is dangerous because it introduces unnecessary functionality with heavier compute, larger input surfaces, and potentially different file/media handling semantics. In this context, the mismatch between declared purpose and embedded capability makes accidental exposure or misuse more likely.

Missing User Warnings

Low
Confidence
77% confidence
Finding
The workflow writes generated outputs to disk without any disclosure in the file, which can create privacy and operational surprises for users who expect transient processing only. In a local-network skill, silent persistence may expose generated content to other local users, backups, or later unintended access.

Static analysis

No suspicious patterns detected.