Surgery-PMPH-10edition

• Outcome: Review
• Latest audit: Jun 4, 2026, 11:34 AM
• Version:
  1.0.0

Overview

This appears to be a medical surgery reference pack rather than malware, but it needs Review because it gives high-risk clinical, drug, emergency, and invasive-procedure instructions without consistent professional-use safeguards.

Treat this as clinician-facing educational reference material only. Do not use it for self-treatment, unsupervised procedures, emergency care, prescribing, or final clinical decisions. Before installing, consider asking the publisher to add prominent professional-use disclaimers, current guideline references, specialist escalation criteria, and to fix the noted internal contradictions.

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 (52)

Intent-Code Divergence

Medium

• Confidence: 94% confidence
• Finding: The skill contains inconsistent timing criteria: the front matter says it applies to patients presenting within 48 hours, while later text states it may be used if onset is not more than 72 hours and lists >72 hours as a contraindication. In a high-risk reduction procedure, conflicting eligibility criteria can cause inappropriate use outside the intended treatment window, increasing the chance of avascular necrosis, missed fracture-dislocation complexity, or delayed specialist care.

Intent-Code Divergence

High

• Confidence: 95% confidence
• Finding: The skill forbids diagnostic puncture/biopsy because cyst-fluid leakage can trigger anaphylaxis or dissemination, yet the operative method immediately recommends puncturing and aspirating the cyst. In a clinical decision skill, this internal contradiction is dangerous because an agent or clinician may generalize the prohibition incorrectly or follow the puncture step without the necessary distinction between prohibited diagnostic needle biopsy and controlled intraoperative decompression under thoracic surgical precautions.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill gives detailed instructions for diagnosing and managing abdominal compartment syndrome, including invasive procedures such as Foley catheter bladder pressure measurement, percutaneous drainage, mechanical ventilation, and decompressive laparotomy, but it does not explicitly warn that this content is only for qualified clinicians in appropriate clinical settings. Because ACS is life-threatening and the interventions can themselves cause serious harm if misapplied, omission of clinician-only/use-context restrictions materially increases the risk of unsafe use by unqualified users or misuse outside supervised care.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: This skill gives concrete diagnostic and antibiotic treatment instructions for a medical condition, including drug classes, treatment duration, and adjunctive measures, but it lacks any safety boundary stating that these recommendations require licensed clinician oversight. In an agent setting, users may follow the guidance directly, creating risk of misdiagnosis, inappropriate antibiotic selection, missed pyelonephritis or STI, adverse drug effects, and antimicrobial resistance.

Natural-Language Policy Violations

Medium

• Confidence: 93% confidence
• Finding: This skill delivers high-risk medical guidance only in Chinese, with no language selection or confirmation that the user understands the instructions. In a burn resuscitation context, misunderstanding fluid formulas, timing windows, or monitoring guidance can directly cause under-resuscitation or fluid overload, leading to serious patient harm.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: This skill gives step-by-step instructions for a dangerous orthopedic reduction maneuver but does not clearly warn that it must only be performed by qualified clinicians in a controlled setting with anesthesia, imaging, and post-reduction neurovascular assessment. Without an explicit user-facing restriction, an unqualified user could attempt the maneuver and cause fractures, sciatic nerve injury, vascular injury, or failed reduction.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: This skill gives actionable bedside guidance that could directly influence debridement or preservation decisions in acute hand trauma, including trimming skin edges and considering excision, without any explicit warning that it is educational information and not a substitute for urgent hand surgery or emergency specialist assessment. In this context, an incorrect or oversimplified viability assessment can cause permanent tissue loss, infection, missed ischemia, or loss of function, making the omission materially dangerous.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: This skill provides prescriptive anti-tuberculosis treatment guidance, including when to start therapy, drug combinations, duration, adverse-effect handling, and cure criteria, but it does not clearly state that the content is for clinician reference only and must not replace individualized medical judgment. Because the topic involves prolonged multidrug therapy with significant toxicity, resistance implications, and special populations such as children and drug-resistant cases, omission of explicit medical-safety disclaimers increases the risk of unsafe or inappropriate use.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill gives concrete hemodynamic treatment instructions for hypotensive or shock patients, including rapid fluid resuscitation, vasoactive choices, and statements such as '严禁盲目补液', but it does not clearly state that it is only decision support and must be used by qualified clinicians with full clinical assessment. In this context, omission of explicit clinician-oversight and safety-boundary language increases the chance that users may over-trust a simplified BP/CVP decision tree and apply unsafe treatment in critically ill patients.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: This skill gives step-by-step instructions for an invasive surgical drainage procedure, including incision design, cavity exploration, and postoperative management, without explicit warnings that it should only be performed by licensed clinicians in an appropriate medical setting. Because the content is actionable and procedural, an unqualified or insufficiently supervised user could attempt a high-risk intervention, causing hemorrhage, duct injury, fistula, worsening infection, or delayed definitive care.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: This skill gives patient-specific cancer subtype classification and corresponding treatment recommendations, including drug classes and regimen components, without an explicit statement that the output is informational only and must be reviewed by a qualified oncologist. In a high-risk clinical context, users may over-rely on the generated recommendation despite omitted factors such as stage, comorbidities, contraindications, pathology nuances, pregnancy status, cardiac function, and evolving guideline requirements.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: This skill provides medical burn-depth classification that is explicitly used to guide treatment and prognosis, but it does not warn that the output is not a substitute for clinician assessment or urgent care. In a high-risk medical context, users may over-rely on the classification, delay emergency evaluation, or make unsafe treatment decisions, especially because burn depth can evolve over the first 72 hours and the skill itself acknowledges dynamic reassessment is necessary.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger examples are broad enough to overlap with general health questions, which can cause the skill to activate outside its intended scope. In a medical skill, over-triggering is more dangerous than usual because it may steer routine wellness queries into cancer-prevention workflows and produce overly specific or risk-stratified advice without sufficient clinical context.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: This skill gives step-by-step instructions for manual reduction and casting of a distal radius fracture, which is a high-risk clinical procedure, but it does not clearly restrict use to qualified clinicians or require pre/post-reduction imaging and documented neurovascular assessment. In this context, omission of those safeguards could enable unsafe use by unqualified users or incomplete clinical evaluation, leading to malreduction, compartment issues, or neurovascular injury.

Missing User Warnings

Medium

• Confidence: 94% confidence
• Finding: The skill gives procedural selection guidance for invasive endoscopic interventions such as EMR/ESD, dilation, stenting, hemostasis, ERCP, and EST, but does not include explicit warnings about major risks like perforation, aspiration, bleeding, pancreatitis, sedation complications, or the need for trained specialists and informed consent. In a clinical skill registry, omission of these safety constraints can enable unsafe or overconfident use, especially if the content is surfaced to non-expert operators or used without adequate supervision.

Missing User Warnings

High

• Confidence: 98% confidence
• Finding: This skill provides step-by-step guidance for endotracheal intubation, a high-risk invasive procedure, but does not include an explicit warning that it is for trained clinicians only, not for laypersons, and not a substitute for supervision, local protocols, or emergency airway expertise. Because the content covers indications, insertion route, depth, and confirmation methods, misuse by an unqualified user could lead to hypoxia, esophageal intubation, aspiration, airway trauma, or death.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill gives directive medical management recommendations, including prophylactic cholecystectomy, without an explicit warning that this is decision support only and must not replace clinician judgment. In a clinical or patient-facing context, users could over-trust the output and act on incomplete criteria, leading to unnecessary surgery, delayed diagnosis, or harm from missed contraindications and individualized risk factors.

Missing User Warnings

High

• Confidence: 97% confidence
• Finding: The skill gives step-by-step instructions for emergency liver trauma surgery, including laparotomy, Pringle maneuver, caval occlusion, and hepatic mobilization, without any explicit restriction that it is for trained surgeons in an appropriate operating-room setting. Because the content instructs immediate invasive action in a high-acuity scenario, omission of professional-use and emergency-escalation warnings could enable unsafe reliance by unqualified users or use outside proper clinical supervision.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: The skill gives decisive, irreversible surgical guidance ('立即切除移植物') in a high-risk transplant context without an explicit warning that this is specialist-only medical content and not suitable for unsupervised use. Even if medically aligned with textbook teaching, presenting it as an operational step without strong qualification, escalation, and context boundaries could cause catastrophic patient harm if applied by non-specialists or in a misdiagnosed case.

Missing User Warnings

High

• Confidence: 95% confidence
• Finding: This index advertises a large catalog of high-risk medical and surgical interventions, including emergency airway management, defibrillation, transfusion, anesthesia, and operative procedures, but does not prominently state that the content is for trained clinicians only and not for layperson use. In an agent setting, that omission increases the chance of unsafe reliance, misuse outside proper supervision, or overconfident execution of dangerous steps with direct patient-harm potential.

Vague Triggers

Medium

• Confidence: 86% confidence
• Finding: The trigger condition is broad and prescriptive: any reducible groin mass that varies with standing or straining causes the skill to be 'immediately' activated, without first excluding important alternatives such as femoral hernia, hydrocele, lymphadenopathy, vascular lesions, or urgent incarceration/strangulation scenarios. In a surgical clinical skill, this can bias assessment toward a narrow hernia differential and promote premature management decisions, increasing diagnostic and patient-safety risk.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill recommends invasive and high-risk clinical actions such as EUS-guided biopsy and operative exploration/ intraoperative ultrasound without any explicit warning about procedural risks, informed consent, contraindications, or the need for qualified specialist supervision. In a medical skill that could influence real-world care decisions, omission of these safeguards can encourage unsafe use by non-specialists or outside appropriate clinical settings.

Missing User Warnings

High

• Confidence: 96% confidence
• Finding: This skill gives step-by-step instructions for invasive medical procedures, including surgical excision, needle aspiration with drug instillation, curettage, and incision/drainage, but does not clearly state that it is intended only for licensed clinicians and must not be used for self-treatment. Even though the metadata says it should be used only after diagnosis and systemic therapy, the operational content is specific enough that an unqualified user could attempt harmful interventions, leading to nerve injury, infection, delayed proper care, or worsening tuberculosis spread.

Missing User Warnings

Medium

• Confidence: 92% confidence
• Finding: The skill instructs users to perform stress and instability maneuvers on an acutely injured knee, which can worsen pain or injury if done by untrained users or without supervision. Although it briefly says actions should be gentle, it lacks explicit warnings to stop for severe pain, avoid self-testing, and restrict use to qualified clinicians, making misuse plausible in a high-risk medical context.

Missing User Warnings

Medium

• Confidence: 95% confidence
• Finding: The skill gives actionable emergency medical guidance for suspected recurrent laryngeal nerve injury, including airway assessment and possible tracheotomy preparation, but does not clearly warn that it is informational only and not a substitute for urgent clinician judgment or emergency response. In a postoperative airway-compromise scenario, omission of that warning can encourage overreliance on the skill, delay escalation, or inappropriate self-directed management in a time-critical situation.

VirusTotal

64/64 vendors flagged this skill as clean.