ClawHub
Back to skill

Security audit

huawei-cloud-msmodelslim-model-analysis

Security checks across malware telemetry and agentic risk

Overview

This model-analysis skill is mostly coherent, but it includes under-warned guidance that could execute third-party model code.

Review before installing if you analyze untrusted Hugging Face or ModelScope models. Only enable `trust_remote_code=True` for trusted, pinned, and reviewed model repositories, preferably inside an isolated environment. The skill otherwise appears to be documentation-only model-analysis guidance with no persistence or credential behavior.

SkillSpector

By NVIDIA
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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)

Vague Triggers

Medium
Confidence
93% confidence
Finding
The trigger list contains broad terms like "model analysis," "adapter," and "transformers" that can match many unrelated requests, causing this skill to activate outside its intended Huawei Cloud msModelSlim pre-adaptation scope. Over-broad activation increases the chance the agent applies specialized workflow instructions to the wrong task, leading to misrouting, incorrect advice, or unnecessary tool use on user-supplied model paths and repositories.

Vague Triggers

Low
Confidence
81% confidence
Finding
The example user phrases include generic requests such as "Check if this model can be quantized" and "Analyze model structure for quantization" without clearly constraining them to supported model families or the msModelSlim workflow. This can broaden routing behavior and encourage invocation for unsupported or unrelated quantization tasks, producing misleading analysis or workflow confusion.

Missing User Warnings

Medium
Confidence
95% confidence
Finding
The troubleshooting guidance recommends setting trust_remote_code=True to load a model but does not warn that this allows execution of model-supplied Python code from the repository. In a model-analysis skill, users are especially likely to inspect third-party or untrusted models, so normalizing this setting without explicit risk guidance can lead to accidental code execution on the analyst's machine or CI environment.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.