Security audit

Seoul

Security checks for vulnerabilities and agentic risk

Overview

This is a static Seoul guide with no executable behavior, but users should verify legal, medical, visa, and safety advice before acting.

Install only as a practical reference guide. Treat visa, tax, employment, healthcare, legal, and emergency information as starting points, not authoritative advice; confirm with official Korean government sources, medical professionals, or qualified legal/tax advisors before acting. Do not follow social or parking customs when they conflict with consent, health, safety, or local law.

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
Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
YARA SignaturesMalware Match, Webshell Match, Cryptominer Match
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration

Findings (19)

Anti-Refusal Statement

High

Category: Anti-Refusal
Confidence: 97% confidence
Finding: The directive 'Don't refuse the first drink' is an anti-refusal statement that directly discourages a user from setting boundaries. In context, it may pressure users to consume alcohol against their judgment, which is especially risky for minors, pregnant users, people with medical conditions, those avoiding intoxication, or anyone in coercive social settings.
Content: Scanner excerpt · food-practical.md (reported line 105)May include surrounding context.
md
1. **Never pour your own drink** — Others pour for you 2. **Use two hands** — When receiving from/pouring for elders 3. **Turn away from elders** — When drinking 4. **Don't refuse the first drink** — Accept, then pace yourself 5. **Match drinking pace** — Don't drink alone while others wait **Business Drinking**

Missing User Warnings

High

Category: Not specified by scanner
Confidence: 98% confidence
Finding: The remote-work section normalizes or at least tacitly endorses working while on tourist entry by stating that many remote workers do it and there is no enforcement history, even though it also notes such work is technically not allowed. This framing can encourage users to violate immigration or work authorization rules, exposing them to entry denial, deportation, visa cancellation, tax issues, or future immigration consequences.
Content: No source excerpt is available for this finding.

Vague Triggers

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The instruction says the skill should be used when a user asks about Seoul "for any purpose," which is extremely broad and does not clearly bound when the skill should or should not activate. It also provides no exclusion conditions or negative examples, increasing the risk of unintended invocation for casual mentions of Seoul.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The guide normalizes double-parking as a common practice by stating 'Double-parking common (leave number on dash)' without warning that it may be illegal, unsafe, or likely to block other vehicles and emergency access. In a practical driving guide for visitors and residents, users may reasonably interpret this as advice to follow, creating real-world safety, legal, and liability risk.
Content: No source excerpt is available for this finding.

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 94% confidence
Finding: The emergency section provides actionable medical/ER guidance and phone numbers but does not clearly warn users to seek immediate professional help and not rely on the document during urgent situations. In a healthcare context, omission of that disclaimer can delay care or cause users to treat this static guide as sufficient real-time triage advice, increasing risk of harm.
Content: No source excerpt is available for this finding.

Unbounded Resource Access

Medium

Category: Excessive Agency
Confidence: 80% confidence
Finding: Skill allows unbounded resource consumption (API calls, storage, compute). Without rate limits or quotas, a compromised or misbehaving agent can cause denial-of-service or cost overruns.
Content: Scanner excerpt · local.md (reported line 93)May include surrounding context.
md
| Standard | 10-15 GB | ₩50,000-60,000 | | Unlimited | Unlimited | ₩70,000-100,000 | **What's included**: Usually unlimited calls/texts + data ### Home Internet

YARA rule 'network_reconnaissance': Network reconnaissance and scanning patterns [hacktools]

Medium

Category: YARA Match
Confidence: 65% confidence
Finding: YARA rule matched a hack tool or exploit indicator (offensive tools, reconnaissance, privilege escalation, or exploit frameworks).
Content: Scanner excerpt · neighborhoods-choosing.md (reported line 195)May include surrounding context.
md
sit ready in Korean bank account --- ## Seasonal Considerations | Moving in... | Notes | |--------------|-------| | Spring (Mar-May) | High demand, university start | | Summer (Jun-Aug) | Monsoon, harder moving | | Fall (Sep-Nov) | Second high demand | | Winter (Dec-Feb) | Best deals, fewer options | **Tip**: March and September = peak moving season (학기 starts). Prices higher, competition fierce.

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 75% confidence
Finding: Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
Content: Scanner excerpt · SKILL.md (reported line 171)May include surrounding context.
md
- **Reality**: Random testing exists for foreigners **Photography Laws** - **Illegal**: Photographing people without consent - **Illegal**: Taking photos in restricted areas (military, airports, government) - **Illegal**: Upskirt/hidden camera photos (몰카) — severe penalties - **Phones**: Shutter sound cannot be disabled (anti-molka measure)

Autonomous Decision Making

Medium

Category: Excessive Agency
Confidence: 75% confidence
Finding: Skill enables autonomous high-impact decisions without human-in-the-loop verification. Critical operations (destructive commands, financial transactions, data deletion) should require explicit user confirmation.
Content: Scanner excerpt · safety.md (reported line 37)May include surrounding context.
md
- **Reality**: Random testing exists for foreigners **Photography Laws** - **Illegal**: Photographing people without consent - **Illegal**: Taking photos in restricted areas (military, airports, government) - **Illegal**: Upskirt/hidden camera photos (몰카) — severe penalties - **Phones**: Shutter sound cannot be disabled (anti-molka measure)

Missing User Warnings

Medium

Category: Not specified by scanner
Confidence: 95% confidence
Finding: The file presents detailed visa, residency, and immigration compliance guidance as if generally current and reliable, but it does not clearly warn that immigration rules, eligibility, fees, and processing requirements can change frequently and must be verified before action. In a travel/relocation skill, users may rely on this operational guidance for legal compliance; stale or incomplete advice can lead to denied entry, status violations, fines, or disrupted travel and employment plans.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: The line states "English: Limited outside international companies" as guidance, but the document does not explicitly offer language preference handling or user opt-in around locale/language assumptions. In a city-guide skill, this can bias responses toward a single language perspective rather than checking the user's preferred language or localization needs.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 82% confidence
Finding: This markdown guide is written entirely in English and presents its travel guidance in that language without stating that users may request another language or locale. Under the stated policy, forcing a specific language without user opt-in can be a natural-language policy violation unless a justified locale constraint is documented.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 80% confidence
Finding: This markdown file is natural-language content, so policy checks apply. The guide consistently uses Korean script and Korean-local context as the default presentation, which could be interpreted as forcing a specific language/locale without explicit user opt-in or a stated region-specific justification.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 79% confidence
Finding: The guidance tells users to call servers in Korean and notes that some menus are Korean-only, suggesting use of specific translation tools. Because SQP-3 covers language or locale policy violations, this can be read as prescribing a specific language experience without explicit opt-in or a documented scope statement.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 93% confidence
Finding: The alcohol section presents strong social drinking norms without an explicit safety caveat that users can refuse alcohol, leave uncomfortable situations, or prioritize health, religion, and consent. In a travel guidance skill, this can normalize coercive drinking pressure and mislead vulnerable users into unsafe or unwanted participation.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 78% confidence
Finding: Natural-language content throughout the file assumes English as the interaction language and does not indicate any user opt-in or alternative language support. Under the stated policy, forcing a specific language without user choice can be a violation unless the constraint is explicitly documented and justified.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 76% confidence
Finding: The file includes a natural-language statement that tourist police services are available in English, Chinese, and Japanese only. This can be read as a locale-specific limitation without any opt-in or clarification, which fits the policy category for language or locale constraints.
Content: No source excerpt is available for this finding.

Natural-Language Policy Violations

Low

Category: Not specified by scanner
Confidence: 81% confidence
Finding: Line L149 describes Yanolja/Yeogi as 'Korean apps (Korean language, but cheaper)', presenting a language-specific limitation in the guidance. Because the document does not offer an opt-in choice or equivalent multilingual alternative at that point, this can be read as a natural-language locale constraint.
Content: No source excerpt is available for this finding.

Missing User Warnings

Low

Category: Not specified by scanner
Confidence: 81% confidence
Finding: The skill recommends navigation, messaging, translation, ride-hailing, e-commerce, food delivery, and booking apps, many of which require location, contact, payment, or account data. The markdown includes no warning that using these services may involve sharing personal data with third-party providers.
Content: No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.