Security audit

Hong Kong

Security checks for vulnerabilities and agentic risk

Overview

This is a content-only Hong Kong guide with no code execution or persistence, though users should verify time-sensitive medical, legal, visa, weather, and tax details before acting.

Install risk is low from an agent-security perspective: this skill is static markdown guidance and does not run commands or access private data. Treat its time-sensitive advice as a starting point only, especially for immigration status, taxes, healthcare, severe weather, short-term rentals, transport legality, and political/legal safety; verify with official Hong Kong sources or qualified professionals before making consequential decisions.

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
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
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

Findings (19)

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 80% confidence
Finding: Skill instructs the agent to omit warnings, disclaimers, or ethical commentary. Stripping safety caveats hides risk from the user and is a common jailbreak preamble.
Content: Scanner excerpt · culture.md (reported line 60)May include surrounding context.
md
2. **Hierarchy:** Wait to be introduced unless you're the senior person 3. **Business cards:** Present and receive with BOTH hands. Study the card before putting it away (never in your back pocket) 4. **Names:** Use title + surname until invited otherwise. Many locals have Western names-use whichever they introduce themselves with 5. **Punctuality:** Expected. Being late without warning = losing face for everyone ### Dining Business

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The activation condition is very broad: 'User asks about Hong Kong for any purpose' can cause the skill to trigger on a wide range of loosely related prompts. This can lead to unintended invocation, overshadowing more specific skills or injecting location-specific guidance when the user did not actually intend to use this skill.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: This file provides actionable healthcare guidance, emergency instructions, medication access information, and cost-based decision advice without a clear disclaimer that it is general informational content and not medical advice. Users may rely on it in urgent or clinical situations, especially because it presents specific recommendations and emergency pathways with an authoritative tone, which increases the risk of harm if details are outdated, incomplete, or not appropriate to the user's condition.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

Category: Not specified by scanner
Confidence: 87% confidence
Finding: SQP-3 covers natural-language policy violations such as forcing a behavior or locale constraint without opt-in. These lines direct users to suppress political discussion and private expression as general guidance, which is a policy-like restriction rather than a neutral safety description.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The file gives detailed visa eligibility, process, fees, timelines, and recommendations, but it does not prominently warn users that immigration rules change frequently and must be verified against official government sources or with a qualified immigration professional. Because users may rely on outdated or incomplete guidance for legal status decisions, this can lead to rejected applications, unlawful work or stay, financial loss, or other immigration consequences.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The typhoon section says typhoons are 'rarely dangerous for tourists,' which can understate a real weather hazard and may encourage users to take storm risks less seriously. In a travel guidance skill, inaccurate minimization of severe weather risk can lead to unsafe decisions such as traveling during warnings or failing to prepare for disruptions and danger.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The file gives specific tax treatment and business-formation guidance in a domain where rules, fees, and eligibility can change, but it does not place a clear warning near the advice telling users to verify with official sources or qualified professionals. In a business/legal/tax context, outdated or overgeneralized statements can lead users to structure operations incorrectly, misreport taxes, or rely on invalid assumptions such as offshore income always being untaxed.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 78% confidence
Finding: This markdown file presents all user-facing content only in English and does not mention any option for alternative languages or user language preference. Under the natural-language policy rule, forcing a specific language without user opt-in can be a policy concern unless explicitly justified.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 81% confidence
Finding: This markdown file is entirely tailored to Hong Kong, including currency, taxes, healthcare, and housing assumptions, but it does not explicitly frame itself as a region-specific guide beyond the title. Under the language/locale policy, locale constraints should be clearly documented and justified; here the scope is implied rather than explicitly stated as a regional limitation for intended use.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 78% confidence
Finding: SQP-3 applies to all file types and covers language or locale policy violations in natural-language content. This section prescriptively centers one language for user interaction and adaptation without any opt-in or explicit acknowledgment that users may need different language accommodations depending on their situation.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The text states that driving tests are in English or Chinese, which imposes a specific language/locale constraint in natural language. Because the document does not frame this as a user choice or provide any opt-in context, it may conflict with the language/locale policy criteria.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 76% confidence
Finding: Line L095 explicitly notes a platform is available in Chinese and English, but the document as a whole assumes this locale context and does not offer any language preference or opt-in. Because SQP-3 covers language/locale policy concerns across all file types, this is a mild policy issue rather than a security flaw.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 84% confidence
Finding: Line L412 states that WhatsApp is 'essential' and that 'everyone uses it for everything,' which is a natural-language recommendation that effectively pressures users toward a specific platform without opt-in or alternatives. Under the policy rule, forcing or strongly imposing a specific communication tool can be a language/locale-style organizational policy concern when user choice is not acknowledged.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 84% confidence
Finding: The guide frames 'English widely spoken' as an advantage of expat enclaves and 'Language barriers' as a disadvantage of local neighborhoods, which embeds an English-default perspective. Because there is no explicit user choice or acknowledgement of language preference, this can be read as a locale-policy issue in natural-language guidance.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The overview repeatedly frames Kowloon in terms of what 'expats' would value, which can amount to a locale or audience-policy issue when the content assumes a specific user demographic by default. Because the file does not indicate that this is intentionally an expat-specific guide or offer alternate framing for other audiences, it risks violating the requirement to avoid forcing a specific locale/audience perspective without opt-in.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 76% confidence
Finding: The line frames language exchange as 'you teach English, learn Cantonese,' which assumes the user is an English speaker and promotes a specific language pairing. Because SQP-3 covers language or locale policy concerns, this is a mild policy issue where the content could be more inclusive by presenting language options rather than a fixed default.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 75% confidence
Finding: This markdown file is entirely tailored to a specific locale, but it does not explicitly state that the guidance is intended only for users interested in Hong Kong. Under the policy rule for natural-language locale constraints, location-specific guidance should be clearly documented and justified rather than implied as a default.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 79% confidence
Finding: The document presents all guidance in English and explicitly normalizes prices to USD via a fixed exchange-rate note, which can be interpreted as imposing a specific language/locale perspective. There is no indication that users can opt into another language or locale format.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 90% confidence
Finding: The lodging guide states Airbnb operates in a legal gray zone and notes some practical risks, but it does not clearly instruct users to verify whether a specific listing complies with Hong Kong licensing rules or platform/local law before booking. In a travel guidance skill, this can normalize potentially non-compliant rentals and expose users to cancellations, denied entry, fines, or lack of recourse.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.