Physiology-PMPH-10edition

• Outcome: Review
• Latest audit: Jun 6, 2026 at 4:39 PM UTC
• Version:
  1.0.0

Overview

This is a coherent physiology education skill pack, but several included skills give direct clinical or procedural guidance without enough local safety limits.

Install only if you want an educational physiology reference and will treat clinical outputs as study material or clinician-review drafts. Do not rely on it for diagnosis, treatment, emergency triage, fertility timing, medication dosing, or unsupervised clinical tests; verify any medical recommendation with qualified clinicians and current guidelines. Be aware that rendered README pages may load third-party images.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access

Findings (21)

Intent-Code Divergence

Medium

• Confidence: 88% confidence
• Finding: The skill’s description says it is suitable for distinguishing leptin deficiency from leptin resistance, but the usage guidance later says it is not applicable to leptin resistance states. This contradiction can mislead users into applying the skill outside its valid scope, producing incorrect biomedical reasoning or treatment expectations in a health-related context.

Intent-Code Divergence

Medium

• Confidence: 93% confidence
• Finding: The skill contains an internal contradiction: it instructs users to evaluate abnormal states such as pneumothorax, then states it is not applicable to pneumothorax and severe chest wall injury. In a medical decision-support context, this ambiguity can cause users to apply the skill beyond its safe scope or misinterpret urgent pathology, increasing the risk of delayed or incorrect clinical action.

Natural-Language Policy Violations

Medium

• Confidence: 93% confidence
• Finding: The skill is entirely in Chinese and does not offer a language choice, which can cause users or upstream agents to misinterpret medical instructions, thresholds, or limitations. In a clinical context, language mismatch increases the chance of incorrect blood gas interpretation or unsafe downstream decision-making, especially for users expecting English or multilingual support.

Vague Triggers

Medium

• Confidence: 90% confidence
• Finding: The trigger condition is broad enough that the skill could be invoked whenever any inhibitory phenomenon is observed, without sufficient gating on context, user intent, or required inputs. This can cause the agent to activate the skill in inappropriate scenarios and produce overconfident classification or intervention guidance outside its intended scope, which is riskier here because the skill includes biomedical decision logic and treatment-oriented suggestions.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: This skill provides clinically significant assessment logic for cerebral perfusion, ischemia, and cerebral edema risk without an explicit user-facing warning that it is informational only and not a substitute for professional medical judgment or emergency evaluation. In a medical context, users may over-rely on the output for triage or treatment decisions, creating risk of delayed care or harmful action when pressure thresholds are oversimplified or applied outside the stated assumptions.

Natural-Language Policy Violations

Medium

• Confidence: 96% confidence
• Finding: The skill explicitly requires a fixed political/ideological framing, including mandated incorporation of specific state ideology, without indicating user choice, contextual necessity checks, or the ability to decline that framing. In an educational content authoring skill, this can bias outputs, reduce neutrality, and pressure downstream users to reproduce ideological content even when their task is pedagogical, scientific, or evaluative rather than political.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill provides clinical screening guidance for EEG interpretation, including references to epilepsy, brain tumors, and pathological abnormalities, but does not warn that it is not a substitute for professional diagnosis. In a medical context, this omission can lead users to over-trust the output, delay care, or self-interpret serious neurological conditions without qualified review.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: This skill gives deterministic medical timing guidance for embryo transfer and endometrial receptivity without any warning that it is not medical advice or that reproductive endocrinologist oversight is required. In an IVF context, users may rely on the output to make treatment-timing decisions, and oversimplified rules about cycle days and synchronization can contribute to mistimed transfer, failed implantation, or inappropriate self-management instead of proper clinical evaluation.

Vague Triggers

Medium

• Confidence: 88% confidence
• Finding: The trigger conditions are broad enough to activate on common medical topics such as anemia, hypoxia, or treatment consideration without clear scoping or user confirmation. In a clinical skill, this can cause the agent to enter a narrow medical-decision workflow prematurely, increasing the chance of overconfident or miscontextualized health guidance.

Natural-Language Policy Violations

Low

• Confidence: 91% confidence
• Finding: The skill is written entirely in Chinese and presents its title, description, usage context, and execution steps only in that language, which can steer the assistant to respond in Chinese even when the user did not request it. This is not a code-execution or data-exfiltration issue, but it can override user preference, reduce usability, and create prompt-behavior inconsistency in multilingual environments.

Natural-Language Policy Violations

Medium

• Confidence: 94% confidence
• Finding: The skill metadata and body are entirely in Chinese and do not provide any mechanism to honor the user's preferred language. In a multilingual agent environment, this can cause the agent to respond in an unintended language, reducing usability, increasing misunderstanding risk, and potentially causing incorrect interpretation of health-related educational content.

Missing User Warnings

Medium

• Confidence: 97% confidence
• Finding: This skill gives clinical guidance about pregnancy status, placental function, risk of preterm birth, and labor initiation without any warning that it is informational only or that abnormal results require evaluation by a qualified clinician. In a high-stakes medical context, users may rely on the output for self-triage or delay urgent care, which can endanger both pregnant patients and fetuses.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill gives concrete treatment recommendations, including a specific drug and dosing example, but does not clearly state that the output is informational only and must not replace clinician judgment. In a medical context, users may over-trust the guidance and apply it without appropriate evaluation of contraindications, patient-specific factors, or regulatory considerations, increasing the risk of harmful self-treatment or unsafe clinical use.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill provides intervention-oriented medical guidance such as monitoring pulmonary artery pressure, giving oxygen therapy, and considering vasodilators, but it does not clearly warn that the content is informational and not a substitute for licensed medical judgment. In a medical context involving hypoxia, pulmonary hypertension, or high-altitude pulmonary edema, users may act on incomplete advice or delay urgent care, which can result in serious harm.

Natural-Language Policy Violations

Medium

• Confidence: 94% confidence
• Finding: The skill is entirely authored in Chinese and does not provide any mechanism to respect or negotiate the user's preferred language. In a multilingual agent environment, this can cause misleading, inaccessible, or non-consensual responses, reducing usability and potentially causing misunderstanding of scientific content.

Missing User Warnings

Medium

• Confidence: 96% confidence
• Finding: The skill gives clinical-style rehabilitation guidance such as prosthetic sensory feedback training, sensory substitution device use, and targeted sensory discrimination exercises, but it does not warn that the output is informational only and not a substitute for qualified medical evaluation. In a medical or rehab context, users may over-trust the guidance and apply it to real patients or themselves despite contraindications, timing issues, or condition-specific risks.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: The skill gives directive medical instructions to immediately start thyroid hormone replacement and pursue endocrine testing without any warning that the content is informational only and not a substitute for clinician evaluation. In a neonatal and infant context, prescriptive guidance can be acted on by non-clinicians or over-trusted by an agent, creating risk of unsafe treatment decisions, misdiagnosis, or delayed emergency care.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill gives step-by-step instructions for a vestibular rotation test that can provoke dizziness, falls, nausea, vomiting, or worsen underlying vestibular, neurologic, or cervical conditions, but it omits warnings, contraindications, and the need for trained clinical supervision. In a clinical-assessment skill, this omission is especially dangerous because users may interpret the procedure as safe and complete guidance for real-world execution.

External Transmission

Medium

• Category: Data Exfiltration

Content

<a href="https://www.star-history.com/#xllgreen/Physiology-PMPH-10edition&Date">
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date&theme=dark" />
   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
 </picture>

• Confidence: 83% confidence
• Finding: https://api.star-history.com/

External Transmission

Medium

• Category: Data Exfiltration

Content

<a href="https://www.star-history.com/#xllgreen/Physiology-PMPH-10edition&Date">
 <picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date&theme=dark" />
   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
 </picture>
</a>

• Confidence: 83% confidence
• Finding: https://api.star-history.com/

External Transmission

Medium

• Category: Data Exfiltration

Content

<picture>
   <source media="(prefers-color-scheme: dark)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date&theme=dark" />
   <source media="(prefers-color-scheme: light)" srcset="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
   <img alt="Star History Chart" src="https://api.star-history.com/svg?repos=xllgreen/Physiology-PMPH-10edition&type=Date" />
 </picture>
</a>

• Confidence: 84% confidence
• Finding: https://api.star-history.com/

VirusTotal

VirusTotal findings are pending for this skill version.