Security audit

Comfyui Painter

Security checks for vulnerabilities and agentic risk

Overview

This local image-generation skill is purpose-aligned, but it needs review because it can use a CivitAI credential, download large models, change local config, and add explicit sexual prompt terms without clear opt-in.

Review before installing. Use it only if you are comfortable allowing an agent to run local ComfyUI commands, access a CivitAI API key, download large model files, and modify local config.json. Require explicit confirmation for downloads and config changes, and remove or constrain the explicit prompt-expansion guidance unless adult-content generation is intentionally enabled and policy-compliant.

SkillSpector

Vulnerability Patterns

Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
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 (3)

Missing User Warnings

Medium

Confidence: 84% confidence
Finding: The skill describes downloading potentially very large model files and automatically updating config.json, but it does not require an explicit warning, confirmation, destination review, or rollback plan before persistent local changes. In an agent context with exec capability, this can lead to unintended disk consumption, overwriting user expectations, and silent configuration drift.

Natural-Language Policy Violations

High

Confidence: 94% confidence
Finding: The prompt optimization guidance hard-codes sexually explicit tags and encourages the agent to transform ordinary user descriptions into explicit sexual content without opt-in or context checks. In a generative media skill, this is dangerous because it can produce disallowed adult content, surprise users, and increase the risk of abusive or policy-violating outputs.

Natural-Language Policy Violations

High

Confidence: 95% confidence
Finding: The supplemental guidance repeats the same explicit sexual tags, reinforcing automatic unsafe prompt expansion as normal behavior. Repetition in the skill instructions makes misuse more likely and increases the chance that an agent will generate explicit content even when the user did not request it directly.

Static analysis

No suspicious patterns detected.