Security audit

Seattle

Security checks for vulnerabilities and agentic risk

Overview

This is a Markdown-only Seattle guide with no executable code or special access, though some neighborhood and social descriptions should be treated as subjective and potentially dated.

Installers should expect a static Seattle reference skill. Use current sources for prices, transit schedules, safety, laws, and business details, and treat broad neighborhood or cultural characterizations as subjective rather than definitive.

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

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
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 (7)

Vague Triggers

Medium

Confidence: 92% confidence
Finding: The activation rule is overly broad: 'User asks about Seattle for any purpose' can cause the skill to trigger on many generic Seattle-related queries, even when a more specific or safer skill would be a better fit. This can lead to inappropriate routing, irrelevant guidance, or unnecessary exposure of the skill in contexts it was not intended to handle, though the content here is informational rather than directly dangerous.

Natural-Language Policy Violations

Medium

Confidence: 91% confidence
Finding: The overview uses broad demographic and cultural generalizations such as higher Asian population, safer feeling, and less urban character as blanket distinctions for the Eastside. In a location-guidance skill, this can encode stereotypes and present subjective claims as facts, which may mislead users or create biased recommendations affecting housing and community perceptions.

Natural-Language Policy Violations

Medium

Confidence: 89% confidence
Finding: Describing Bellevue as very international followed by a parenthetical list of ethnic groups ties neighborhood identity to broad ethnicity-based characterizations. While likely intended as helpful context, this can reduce residents to demographic shorthand and may violate neutrality expectations in a consumer-facing guidance skill.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: The clothing guidance states 'Waterproof shell — Not umbrella' and repeats 'Rain jacket (not umbrella),' which imposes a specific behavioral/local-norm preference rather than offering a user choice. This is a natural-language policy issue because it dictates one approach instead of presenting alternatives or making the recommendation optional.

Natural-Language Policy Violations

Low

Confidence: 84% confidence
Finding: Listing Asian immigrants as a defining resident category generalizes who belongs in the area and can reinforce demographic profiling. In this skill context the harm is primarily reputational and fairness-related rather than directly exploitable, but it still risks alienating users and producing biased neighborhood framing.

Natural-Language Policy Violations

Low

Confidence: 77% confidence
Finding: The file describes the International District as "Seattle's Asian neighborhood" and "Asian culture center," which applies a broad ethnic label to a place in a way that can be overly reductive. This is a natural-language policy concern because the phrasing categorizes a community by race/ethnicity without qualification or user-selected framing.

Natural-Language Policy Violations

Low

Confidence: 80% confidence
Finding: The section on 'The Seattle Freeze' states as fact that 'People are polite but not warm' and that invitations may not be forthcoming, presenting a generalized characterization of a population without qualification. This kind of locale-linked behavioral framing can be interpreted as a language/locale policy issue because it assigns social traits to residents of a place rather than presenting them as subjective or optional context.

Static analysis

No suspicious patterns detected.