Back to skill

Security audit

Morpheus Fashion Design

Security checks across malware telemetry and agentic risk

Overview

The skill appears aimed at AI fashion image generation, but it also sends user assets to ComfyDeploy and appears to run an under-disclosed local usage-tracking script.

Review this before installing if you will use private product photos, model images, client assets, or identifiable people. The ComfyDeploy upload is expected for the advertised workflow, but the package should more clearly disclose that upload and should remove, disable, or make opt-in the local usage-tracking script before sensitive use.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Findings (8)

subprocess module call

Medium
Category
Dangerous Code Execution
Content
"--type", "image",
            "--status", status
        ]
        subprocess.run(cmd, capture_output=True, timeout=10)
    except Exception as e:
        print(f"Warning: Failed to track usage: {e}", file=sys.stderr)
Confidence
91% confidence
Finding
subprocess.run(cmd, capture_output=True, timeout=10)

Lp3

Medium
Category
MCP Least Privilege
Confidence
92% confidence
Finding
The skill documentation includes capabilities to access local files, invoke shell commands, and make network requests, but it declares no permissions or trust boundaries. This can cause agents to perform higher-risk actions than users expect, especially reading from local paths and transmitting data externally without explicit authorization controls.

Context-Inappropriate Capability

Medium
Confidence
88% confidence
Finding
The skill instructs the agent to search a local model catalog via shell commands even though the core function is image generation through a remote API. This expands the attack surface to local filesystem and command execution, creating opportunities for unintended data access or abuse of local resources beyond the user’s expected task.

Context-Inappropriate Capability

Medium
Confidence
97% confidence
Finding
The skill performs side-effect tracking through a separate script in the user's home directory, which is outside the core image-generation function and not disclosed in the skill description. This can leak prompts, output paths, and status metadata to another system without informed consent and expands the attack surface through a secondary executable.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The examples direct the agent to send product and model image URLs to a third-party service, but the skill does not provide a clear user-facing disclosure that these images will leave the local environment. Because model images may contain personal data, silent transmission can violate user expectations, privacy requirements, or organizational policy.

Missing User Warnings

Medium
Confidence
93% confidence
Finding
The local catalog guidance instructs selection and filtering based on ethnicity, gender, and body type without any warning about sensitive attribute handling or fairness/privacy safeguards. This encourages processing of demographic data in a way that may be restricted, biased, or non-compliant depending on jurisdiction and use case.

Missing User Warnings

Medium
Confidence
98% confidence
Finding
The script uploads product images, model images, optional logos, and text prompts to a third-party API and may also forward prompt/result metadata to a tracker, yet it provides no explicit privacy warning or consent gate. In this skill context, those assets can be commercially sensitive, personally identifiable, or subject to contractual restrictions, so silent transmission materially increases privacy and compliance risk.

External Transmission

Medium
Category
Data Exfiltration
Content
p = Path(filepath)
    mime = "image/png" if p.suffix == ".png" else "image/jpeg"
    with open(p, "rb") as f:
        r = requests.post(
            "https://api.comfydeploy.com/api/file/upload",
            headers={"Authorization": f"Bearer {api_key}"},
            files={"file": (p.name, f, mime)},
Confidence
94% confidence
Finding
requests.post( "https://

VirusTotal

66/66 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.