Security audit

huxishu

Security checks across malware telemetry and agentic risk

Overview

This is a coherent TCM reference skill with no executable code, but it gives actionable diagnosis, formula, and herb dosage guidance without clear medical safety boundaries.

Review before installing if this will be available to general users. It should be treated as historical or educational TCM material, not a self-diagnosis or self-treatment tool. Add explicit medical boundaries, emergency and red-flag escalation, clinician-supervision requirements for formulas and dosing, and warnings against self-administering herbs based on agent output.

SkillSpector

By NVIDIA

Vulnerability Patterns

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep

Findings (5)

Missing User Warnings

High

Confidence: 97% confidence
Finding: This section gives concrete herbal dosing guidance, conversion formulas, and administration heuristics without any visible safety disclaimer, clinician oversight requirement, or contraindication framework. Users could self-administer potent substances such as附子、麻黄、细辛、大黄 based on the text alone, creating real risk of toxicity, delayed care, or dangerous misuse.

Missing User Warnings

High

Confidence: 98% confidence
Finding: These sections map symptoms and syndromes directly to treatments or formulas (for example, '有汗用桂枝，无汗用麻黄' and similar condition-to-remedy rules) without warning that such simplifications can be unsafe outside professional evaluation. This can encourage users to self-diagnose and self-treat potentially serious conditions, including missing emergencies or using inappropriate remedies.

Missing User Warnings

High

Confidence: 97% confidence
Finding: The file provides concrete diagnosis and treatment-selection rules, named formulas, escalation logic, prognostic markers, and medication adjustment guidance without any safety guardrails, uncertainty framing, or instruction to seek licensed medical care. In the context of a user-facing agent skill, this can enable unsafe self-diagnosis or self-treatment, especially because the content is written as decisive heuristics ('即可用柴胡剂', '必须先救里', '主之', '死证五标志') that may be acted on directly.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The file provides concrete medical and quasi-prescriptive treatment guidance, including syndrome interpretation, medication/formula selection, contraindications, and recovery instructions, without any warning that this is historical/educational TCM content rather than professional medical advice. Because it frames symptoms and treatment choices as operational rules, a user could self-diagnose or delay proper care, creating meaningful risk of harm in a health-related skill.

Missing User Warnings

High

Confidence: 97% confidence
Finding: This section gives concrete herb dosage conversions, routine/max doses, and escalation heuristics for potent substances without any safety disclaimer, contraindications, licensing boundary, or instruction to seek qualified medical supervision. In a general-purpose agent skill, users may treat this as actionable prescribing guidance and self-administer unsafe amounts, especially where the text normalizes dose adjustment and questions traditional dosage limits.

VirusTotal

VirusTotal findings are pending for this skill version.

View on VirusTotal

Static analysis

No suspicious patterns detected.