Security audit

Shi Gou

Security checks across malware telemetry and agentic risk

Overview

This security helper is mostly coherent, but its command redaction output can expose the original secrets it is supposed to hide.

Review before installing if you plan to process real credentials. The sanitizer prints a redacted command but also returns the original matched secret values in metadata, so its full JSON output should not be logged, pasted into chats, or shared until that behavior is changed. The broad triggers may also cause accidental activation for unrelated requests.

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: 86% confidence
Finding: The trigger phrase '扫描' is very broad and can match many ordinary requests, causing unintended activation of the security skill in irrelevant contexts. Overbroad activation can disrupt agent routing, override more appropriate skills, and create denial-of-service-like behavior through unnecessary scanning or false alarms.

Vague Triggers

Medium

Confidence: 84% confidence
Finding: The trigger '脱敏' is generic and may activate on unrelated conversations about masking, privacy, editing, or formatting. In an agent ecosystem, this can lead to accidental invocation, data transformation when not requested, and leakage of workflow control to the wrong skill.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The phrase '敏感信息' is too broad to serve as a safe activation trigger because it appears in many benign discussions unrelated to command sanitization. This increases the chance of unintended skill firing, misrouting, and unnecessary processing of user content that may itself contain sensitive data.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

Detected: suspicious.dynamic_code_execution, suspicious.exposed_secret_literal, suspicious.prompt_injection_instructions

Dynamic code execution detected.

Critical

Code: suspicious.dynamic_code_execution
Location: scripts/security-check.js:34

File appears to expose a hardcoded API secret or token.

Critical

Code: suspicious.exposed_secret_literal
Location: README.md:62

Prompt-injection style instruction pattern detected.

Warn

Code: suspicious.prompt_injection_instructions
Location: README.md:29

Prompt-injection style instruction pattern detected.

Warn

Code: suspicious.prompt_injection_instructions
Location: SKILL.md:231