Security audit

Pharmacology-PMPH-10edition

Security checks for vulnerabilities and agentic risk

Overview

This is a pharmacology reference skill with no executable payload, but it gives broad high-risk medical treatment and dosing guidance without consistent clinician-only safeguards.

Review this before installing if you might use it outside a supervised clinical or educational setting. Treat outputs as textbook study material or clinician-reviewed reference only, and do not rely on it for diagnosis, prescribing, dosing, emergency treatment, pregnancy-related medication decisions, chemotherapy, anticoagulation reversal, insulin changes, or poisoning care without qualified medical oversight and current local protocols.

A.I.G

Vulnerability Patterns

Skill Instruction HijackingAlters the agent's session goals or safety constraints when the skill loads
Agent Memory PoisoningWrites attacker-controlled rules into memory that affect later sessions
Remote Payload Retrieval and ExecutionFetches external code whose behavior can change after review
Embedded Malicious CodeShips malicious scripts inside the skill and executes them locally
Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs

SkillSpector

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (242)

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: This README presents a large library of pharmacology and clinical decision-support skills but does not prominently warn that outputs may influence diagnosis, prescribing, dosing, or emergency treatment. In this domain, absence of a medical safety disclaimer and escalation guidance materially increases the risk that users treat the content as authoritative bedside advice, potentially causing patient harm.
Content: No source excerpt is available for this finding.

Ae1

High

Category: analysis-evasion
Confidence: 100% confidence
Finding: Referenced artifact was not completely inspected
Content: Scanner excerpt · SKILL.md (reported line 35)May include surrounding context.
md
本技能包包含 180 个独立技能，每个技能对应一个文件夹，内含 `SKILL.md` 文件。通过技能索引 (`index.md`) 可快速定位所需知识点。

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill provides actionable emergency-toxicology instructions including gastric lavage, IV physostigmine, diazepam use, ventilatory support, and treatment sequencing, but it does not clearly state that these are clinician-only interventions requiring urgent poison-center or emergency care supervision. In a general-purpose agent context, omission of a strong professional-use warning can enable unsafe self-treatment or delayed emergency response, especially because the content concerns life-threatening poisoning and narrow-risk antidote administration.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill gives direct treatment and prophylaxis recommendations for a prescription antiviral without clearly requiring clinician judgment, diagnostic confirmation, and verification against current official prescribing information. Because this is healthcare content, omission of these safeguards can lead to inappropriate autonomous medication guidance, contraindicated use, or outdated recommendations being acted upon.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill provides high-risk medication selection guidance for benzodiazepines, including IV diazepam for status epilepticus and an instruction to output drug name, dose range, route, and treatment course, but does not clearly state that this is clinician-only information and not for self-medication. Because benzodiazepines are controlled sedatives with overdose, respiratory depression, dependence, withdrawal, and misuse risks, omission of strong supervision and emergency-care warnings could enable unsafe or unauthorized use.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill recommends specific pharmacologic strategies for serious neurologic and psychiatric conditions such as Parkinson disease, schizophrenia, and Huntington disease without requiring clinician supervision, dosage review, contraindication screening, or emergency triage. In this context, users could overgeneralize symptom-based pathway guesses into self-medication or inappropriate treatment changes, creating substantial risk of harm from delayed diagnosis, adverse effects, or worsening psychosis/movement disorders.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The skill gives actionable instructions to initiate chlorpromazine, select candidates, define maintenance treatment, and stop therapy, but it does not prominently state that antipsychotic prescribing is high-risk and must only be performed by a qualified clinician with patient-specific assessment and monitoring. Because chlorpromazine can cause serious harms including hypotension, extrapyramidal symptoms, seizures, anticholinergic effects, and treatment misselection, a user could apply this guidance without adequate oversight and cause significant patient harm.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill gives operational guidance for designing cytotoxic chemotherapy regimens, including drug class sequencing based on growth fraction and cell-cycle phase, but does not clearly state that this is high-risk clinician-only content and not suitable for patient self-management. Because chemotherapy has narrow therapeutic margins and serious toxicity, omission of an explicit safety warning could enable unsafe use, overreliance by non-specialists, or harmful clinical decisions without oncology oversight.
Content: No source excerpt is available for this finding.

Vague Triggers

High

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill exposes high-risk clinical treatment guidance using broad activation criteria such as shock type and absence of bleeding tendency, without requiring physician oversight, patient-specific dose limits, diagnostic confirmation, or emergency-care context. In a medication-selection skill involving dextran for shock and antithrombotic use, overly permissive triggering can cause unsafe recommendations in patients with occult bleeding, coagulopathy, renal failure, or fluid overload risk.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill gives condition-based guidance for emergency thrombolysis selection and timing in acute myocardial infarction, pulmonary embolism, and ischemic stroke without an explicit, prominent warning that it is educational only and must not be used as a standalone treatment decision tool. Because fibrinolytics carry major hemorrhage risk and disease-specific eligibility windows and contraindications are more nuanced than presented here, a user could over-rely on this content and make unsafe time-critical decisions without urgent specialist oversight.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The skill gives direct medication-selection guidance for heart failure subtypes, including when positive inotropes may be used or avoided, but it does not include an explicit warning that the output is only supportive and must not replace clinician judgment or independent verification. In a medical context, this can encourage overreliance on simplified rules despite important contraindications, hemodynamic nuance, and patient-specific factors, creating a meaningful risk of harmful treatment decisions.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill gives actionable medication-selection and execution instructions for H. pylori eradication, including specific antibiotics and a fixed 14-day regimen, but omits key safety controls such as clinician oversight, contraindication review, pregnancy considerations, drug-drug interactions, adverse effects, and resistance-guided selection. Because it explicitly says to '制定并执行' the regimen, a user or agent could apply prescription therapy without adequate medical review, creating risk of allergic reactions, treatment failure, C. difficile-associated diarrhea, hepatotoxicity, and inappropriate antibiotic use.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill gives direct instructions to stop heparin immediately and administer protamine with dose ratios, but it does not clearly state that these are urgent, high-risk interventions requiring clinician supervision and patient-specific assessment. In a general agent skill, this can enable unsafe self-management or overconfident non-specialist use, leading to thrombosis from abrupt anticoagulant cessation, under-treatment of major bleeding, or adverse reactions such as protamine-induced hypotension/anaphylaxis.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill gives concrete emergency-treatment instructions for suspected hypoglycemia, including IV 50% glucose for coma, but does not clearly warn that this is not a substitute for urgent professional or emergency care. In a patient-facing or loosely supervised agent context, users may delay calling emergency services, misdiagnose other causes of altered mental status, or attempt unsafe interventions beyond their competence.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 99% confidence
Finding: The skill recommends that acute insulin resistance may require large short-term insulin escalation up to hundreds or thousands of units per day, without explicit safeguards about ICU-level monitoring, specialist oversight, infusion protocols, and frequent glucose/electrolyte checks. If acted on inappropriately, this could cause catastrophic dosing errors, severe hypoglycemia, potassium shifts, and death.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The skill provides actionable insulin-selection and administration guidance for high-risk scenarios such as diabetic ketoacidosis, perioperative care, hyperosmolar coma, and potassium-shift management, but it does not clearly state that the output must not replace clinician judgment, institutional protocols, or urgent bedside assessment. Because insulin is a narrow-therapeutic-index medication with serious risks from dosing, route, and monitoring errors, an LLM using this skill could produce overconfident recommendations that are applied without adequate safeguards.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: The skill provides actionable anesthesia administration selection guidance, including drug choice, epinephrine coadministration, neuraxial techniques, and solution density manipulation, but does not prominently require qualified clinician oversight or warn that misuse can cause catastrophic harm such as local anesthetic systemic toxicity, nerve injury, total spinal anesthesia, or death. In a clinical-decision skill, omission of safety guardrails materially increases the risk that a user may treat it as sufficient procedural guidance and apply it outside an appropriately supervised setting.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: This is high-risk emergency medical guidance that includes drug administration and acute resuscitation steps, but it lacks an explicit warning that it must only be used by qualified clinicians under appropriate protocols. In a real clinical or non-clinical setting, users may over-rely on the skill as a stand-alone instruction source, leading to unsafe dosing, delayed escalation, or inappropriate treatment during a life-threatening local anesthetic systemic toxicity event.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill gives concrete pharmacogenomic drug-selection and dose-adjustment guidance, including avoiding one drug and substituting another, but does not state that the output is clinical decision support only and must be verified by a licensed clinician before therapy changes. In a medication-dosing context, users may over-trust the instructions and make unsafe prescribing or self-medication decisions, especially because genotype interpretation is only one part of safe dosing and must be integrated with indication, comorbidities, organ function, interactions, and current guidelines.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: This is high-risk clinical dosing guidance for primaquine, including a recommendation to double the dose when resistance is suspected, but it does not explicitly require clinician verification, weight/indication-specific confirmation, or adherence to local/national malaria guidelines. In a medical skill, omission of these safeguards can lead users to apply unsafe or outdated regimens, increasing the risk of hemolysis, methemoglobinemia, treatment failure, or inappropriate use in vulnerable patients.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 97% confidence
Finding: This skill gives operational guidance for selecting and administering uterotonic prostaglandins for abortion, induction, molar pregnancy evacuation, and amniotic fluid embolism rescue, but does not explicitly require qualified obstetric supervision or emergency-capable care. Because these drugs can cause severe maternal and fetal harm if misused, omission of a clear safety boundary could enable unsafe self-administration or inappropriate use in non-specialist settings.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 98% confidence
Finding: This skill gives direct treatment-initiation instructions, including route and dose of filgrastim-equivalent therapy, without an explicit warning that it must only support qualified clinician judgment and not serve as a standalone decision-maker. In a medical context, omission of such guardrails can lead to unsafe autonomous use, inappropriate administration, missed differential diagnoses, or failure to account for patient-specific contraindications and monitoring needs.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: The skill gives operational instructions for urinary alkalinization and acidification to alter drug elimination, including concrete targets and agents, but omits critical safeguards such as clinician/toxicologist supervision, contraindications, monitoring of electrolytes/acid-base status, and the fact that urine acidification is generally hazardous and rarely recommended. In the context of poisoning management, incomplete guidance can directly cause patient harm through metabolic derangements, fluid overload, arrhythmia, or delayed use of more appropriate treatments.
Content: No source excerpt is available for this finding.

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 96% confidence
Finding: This content gives clinically actionable guidance about manipulating urine pH to alter drug excretion, including contraindications, but does not explicitly warn that such interventions require clinician supervision and are not appropriate for self-treatment. Because urine alkalinization/acidification can worsen heart failure, renal impairment, liver disease, or acid-base disorders, omission of a clear medical-use warning creates a meaningful risk of unsafe application by end users.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The file’s natural-language interface, description, and operating instructions are entirely in Chinese, with no indication that the user can choose another language. Under the stated policy, forcing a specific language without user opt-in is a natural-language policy violation unless the locale restriction is explicitly justified.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.