Security audit

huawei-cloud-msot-msopprof-operator-profiler

Security checks across malware telemetry and agentic risk

Overview

This skill is mainly a legitimate Ascend NPU profiling guide, but it includes unsafe troubleshooting commands and under-described profiling data persistence that users should review before installing.

Use this only in a controlled Ascend profiling environment. Review commands before running them, avoid disabling SELinux or using chmod 777 unless you fully understand and accept the risk, and confirm where profiling data will be stored before exporting it to any database.

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 (5)

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger list contains generic terms such as "profiler," "performance profiling," and "profile data," which can cause the skill to activate in contexts unrelated to Ascend NPU operator profiling. Over-broad invocation increases the chance the agent runs profiling-oriented guidance or commands in the wrong workflow, leading to unintended tool use or confusion.

Missing User Warnings

Medium

Confidence: 76% confidence
Finding: The workflow explicitly includes exporting collected profiling data to a profiling database, but the skill does not warn users that operational metadata, model characteristics, paths, or other sensitive performance artifacts may be persisted externally. In enterprise environments, silent export behavior can create data-governance and confidentiality risks even if the skill is intended for legitimate profiling.

Missing User Warnings

Medium

Confidence: 98% confidence
Finding: The troubleshooting advice tells users to run `setenforce 0` to disable SELinux enforcement, but provides no warning about the security implications or safer alternatives. Disabling SELinux reduces host protection and can allow otherwise confined processes or compromised tooling to access files, sockets, or devices more broadly, which is especially risky on systems running profiling tools and executable binaries.

Missing User Warnings

Medium

Confidence: 99% confidence
Finding: The guidance recommends `chmod 777 /tmp/profiling`, making the output directory world-writable without any caution. World-writable directories can enable tampering, symlink attacks, or unauthorized modification of profiling outputs, which is dangerous in shared or multi-user environments and unnecessary for normal operation.

Missing User Warnings

Medium

Confidence: 89% confidence
Finding: The verification script includes `rm -rf output_npu` before running profiling, which performs destructive deletion without confirmation or safeguards. While it targets a fixed relative directory rather than an obviously attacker-controlled path, it can still cause unintended data loss if the directory contains prior results or if the script is run from an unexpected working directory.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.