Back to skill

Security audit

Psychiatry-PMPH-9edition

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed psychiatry reference skill, but it includes detailed medication, emergency, procedural, and involuntary-treatment guidance without consistent runtime safeguards limiting use to qualified clinicians.

Install only if you intend to use it as educational or clinician-reviewed psychiatry reference material. Do not rely on it for self-diagnosis, self-medication, emergency care, involuntary treatment decisions, or legal-capacity determinations without qualified professional review and local guideline/legal checks.

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
Vulnerability Patterns
  • 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
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (164)

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

This skill catalog covers high-risk domains such as suicide assessment, violence management, involuntary treatment, overdose response, and psychopharmacology, but it provides no user-facing warning about limitations, clinician oversight, emergencies, or jurisdiction-specific/legal constraints. That omission can lead users to rely on the skill as definitive real-world clinical guidance during crises, creating elevated risk of patient harm, inappropriate coercive actions, or dangerous delay in seeking emergency care.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill gives actionable medication-selection guidance for Alzheimer's disease and BPSD, including psychotropics and anti-dementia drugs, but does not prominently warn that it is intended only for qualified clinicians and not for patient self-medication. In a high-risk medical domain, this can enable unsafe self-treatment, misuse in unqualified settings, and harm from adverse effects, contraindications, or delayed proper evaluation.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The document gives concrete prescription dosing, titration, combination use, and antipsychotic management instructions for a vulnerable population without a clear warning that it is clinician-facing reference material only and not patient self-medication guidance. In a deployed agent, this can enable unsafe medication use, dose escalation, or continuation without individualized assessment, especially given the inclusion of high-risk drugs for elderly dementia patients.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill gives condition-specific diagnostic and treatment guidance, including medication selection, benzodiazepine use, augmentation strategies, and treatment timelines, but does not include visible safety boundaries, triage rules, or a warning that the content is not a substitute for direct clinical evaluation. In an LLM setting, users may act on this as individualized medical advice, which is especially dangerous for psychiatric symptoms where suicidality, psychosis, substance withdrawal, medical mimics, or acute deterioration may require urgent in-person assessment.

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 specific psychiatric assessment and treatment instructions, including medication selection, behavioral therapy choices, and urgent management of medical complications, but it does not warn that this is clinician-facing medical guidance requiring qualified oversight. In a healthcare context, omission of scope, supervision, and safety boundaries can lead to inappropriate self-treatment, misuse by unqualified users, or overreliance on incomplete guidance in children and other vulnerable patients.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The instructions provide structured mental-health relapse management guidance but do not include user-facing warnings about the limits of the skill, nor explicit emergency escalation steps for suicidality, rapid deterioration, psychotic features, or inability to care for self. Because this concerns depression in a recovery phase, omission of safety guardrails is dangerous: relapse can evolve into acute risk, and users may rely on the workflow instead of seeking urgent clinical evaluation.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The document gives operational instructions for hypnosis, IV barbiturate/sedative use, electrostimulation, and MECT in a procedural SOP style without prominent safety warnings, contraindications, monitoring requirements, or an explicit restriction to trained specialists in appropriately equipped settings. In a clinical decision-support skill, this can enable unsafe real-world use by underqualified users or encourage decontextualized application of high-risk interventions, creating risks of respiratory depression, cardiac complications, procedural injury, and other serious harm.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file provides specific prescription drug doses and titration schedules for narcolepsy and idiopathic hypersomnia, including controlled or high-risk agents, but does not include an explicit warning that these medications require physician supervision and should not be self-administered based on the document alone. In a clinical-support skill, omission of guardrails is dangerous because the content can be surfaced to non-specialists or patients, enabling unsafe self-medication, dose escalation, misuse, or omission of contraindication monitoring.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The file provides specific antipsychotic dosing and titration guidance for delirium without an explicit warning that this is clinician-only reference material requiring individualized medical judgment, monitoring, and contraindication review. In a general-purpose agent context, such content could be surfaced to non-clinicians or used without adequate safeguards, creating risk of inappropriate treatment, adverse drug events, or delayed urgent medical evaluation for a potentially life-threatening condition.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The file provides concrete medical management guidance, including blood pressure, glycemic and lipid targets, antiplatelet/anticoagulation use, cognitive drug dosing, antipsychotic use principles, and follow-up intervals, but does not state that it is reference material requiring clinician judgment and patient-specific evaluation. In a safety-critical medical context, omission of this warning can enable overreliance on generalized recommendations despite major contraindications, comorbidities, drug interactions, frailty, renal function, bleeding risk, and diagnostic uncertainty.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

This document provides actionable medical dosing, oxygen therapy thresholds, airway escalation steps, injectable medication regimens, and other high-risk interventions without an explicit warning that it must only be used by qualified clinicians under formal medical supervision. If surfaced directly to end users or used out of context, it could enable unsafe self-treatment, delayed emergency care, medication misuse, or harmful application to patients without individualized assessment.

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 structured diagnostic guidance for ICD-11 personality disorder assessment, severity grading, subtype differentiation, and etiologic interpretation without any explicit limitation statement, safety warning, or requirement for qualified psychiatric oversight. In a mental-health context, unqualified or overconfident use can lead to misdiagnosis, stigma, inappropriate treatment decisions, or failure to escalate risk cases such as self-harm, psychosis, or emergency presentations.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill provides concrete psychiatric drug selection, dosing, titration, toxicity management, and MECT procedural guidance, including emergency interventions, but does not clearly state that it is restricted to qualified clinicians and cannot replace specialist judgment or local protocols. In this context, omission of an explicit safety gate is dangerous because an unqualified or overconfident user could act on high-risk medication and emergency recommendations directly, causing overdose, adverse interactions, delayed escalation, or improper treatment.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

This file provides step-by-step instructions, dosing, device parameters, and complication responses for a high-risk procedure (MECT/ECT with anesthesia and neuromuscular blockade) but does not prominently restrict use to licensed clinicians in appropriately equipped, supervised medical settings. Because the content is operationally detailed, a reader could treat it as actionable guidance outside proper governance, increasing the risk of serious injury or death from misuse, outdated practice, or omission of institutional 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 gives concrete antipsychotic selection, dose-titration, maintenance duration, and combination-treatment guidance in a prescriptive manner, but does not clearly warn that such information requires specialist psychiatric evaluation, individualized risk assessment, and supervision. In a high-risk domain like schizophrenia treatment, omission of those safeguards can lead users to apply medication advice inappropriately, including unsafe polypharmacy, delayed escalation, or failure to account for contraindications and emergencies.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill identifies suicide and violence risk factors but stops at risk description, without explicitly instructing urgent emergency intervention when immediate danger may exist. In psychiatric contexts, presenting safety-risk indicators without a clear escalation path can normalize passive assessment and contribute to delayed crisis response for imminently dangerous patients.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill contains detailed psychiatric/addiction diagnosis, withdrawal timelines, acute intoxication management, and specific medication/dosing instructions for substances such as methadone, buprenorphine, diazepam, haloperidol, and others, but it does not clearly restrict use to licensed clinicians or include a safety warning that this is not for lay self-treatment. In this context, incorrect self-application or use by an underqualified operator could lead to undertreatment, overdose, dangerous withdrawal complications, missed emergencies, or harmful drug interactions, making the absence of explicit safety boundaries a real safety/security issue.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
94% confidence
Finding

The section describes delirium tremens and notes a nontrivial mortality risk, but it does not instruct urgent escalation, emergency referral, or immediate supervised care. In this skill context, omission is dangerous because users may treat severe symptoms as informational only, delaying emergency intervention for a time-critical condition with risk of death, seizures, autonomic instability, and profound confusion.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
96% confidence
Finding

The document gives concrete benzodiazepine dosing and naltrexone follow-up instructions for alcohol withdrawal without an explicit clinician-only restriction, screening prerequisites, contraindications, or warning against self-management. Alcohol withdrawal can rapidly become life-threatening, and simplified dosing guidance may be misused by non-clinicians or applied without monitoring, leading to under-treatment, oversedation, respiratory depression, missed delirium tremens, or delayed emergency care.

Content

No source excerpt is available for this finding.

Missing User Warnings

High
Category
Not specified by scanner
Confidence
95% confidence
Finding

The file provides concrete opioid withdrawal induction timing and dose-escalation instructions for buprenorphine and buprenorphine/naloxone without any visible warning that this is clinician-facing medical content requiring professional supervision, patient assessment, and local protocol compliance. In a general-purpose agent skill, such guidance could be surfaced to non-professionals or applied without screening for opioid type, dependence severity, precipitated withdrawal risk, respiratory compromise, pregnancy, polysubstance use, or regulatory constraints, creating a meaningful risk of patient harm.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
83% confidence
Finding

The main README content is presented in Chinese, while other languages are relegated to alternate files via links. This creates a default language constraint rather than offering an in-place language choice or explicit user opt-in, which may violate language/locale policy expectations.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger guidance is overly broad because it allows activation '按技能描述触发' without explicit trigger phrases, boundaries, or exclusion conditions. In a safety-critical psychiatry skillset, ambiguous activation can cause the assistant to invoke high-risk clinical guidance in unintended contexts, increasing the chance of unsafe or overconfident medical recommendations.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
77% confidence
Finding

The description and all operational guidance are entirely in Chinese, with no indication that the user may request another language or that the skill is intentionally limited to a Chinese-language clinical context. Under the stated policy, forcing a specific language without opt-in is a natural-language policy issue.

Content

No source excerpt is available for this finding.

Ae4

Medium
Category
analysis-evasion
Confidence
80% confidence
Finding

Suspicious Unicode normalization or mixed-script content

Content

No source excerpt is available for this finding.