Security audit

hairstyle-recommender

Security checks for vulnerabilities and agentic risk

Overview

The skill is coherent for hairstyle advice, but it mandates portrait-based image generation without clear user consent or privacy disclosure for external tools.

Install only if you are comfortable with the agent analyzing portrait attributes and potentially using or preparing your portrait for external image-generation services. Users should be told before any image upload or third-party generation step and should have a clear option to receive advice without generating an image.

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

Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
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
Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (4)

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The README states the skill will trigger automatically when a user uploads a portrait photo and asks for hairstyle advice, and the metadata also includes broad trigger phrases. Broad activation on sensitive image uploads can cause unintended invocation and analysis of biometric-like facial attributes without sufficiently explicit user intent. In a photo-analysis skill, over-broad triggering is more dangerous because portraits are highly sensitive inputs.

Missing User Warnings

Medium

Confidence: 95% confidence
Finding: The README describes analyzing portrait photos for face shape, facial features, and hair characteristics, but does not warn users that uploaded portraits will be processed for sensitive appearance analysis. This creates a transparency and consent problem: users may not realize the extent of analysis performed on their image data. Because the skill centers on facial-image processing, missing disclosure is materially risky in context.

Natural-Language Policy Violations

Medium

Confidence: 94% confidence
Finding: The skill hard-requires Chinese output, regardless of the user's language, which can undermine informed consent and safe use when users do not understand the instructions or privacy implications. In this skill, that matters because users may be asked to reuse detailed prompts with external image-generation services and may miss important caveats if they cannot read Chinese.

Missing User Warnings

Medium

Confidence: 97% confidence
Finding: The skill instructs the assistant to generate or provide prompts for third-party image-generation platforms using the user's portrait, but it does not give an upfront privacy warning that the photo or derived facial data may be sent to external services. Because this skill processes sensitive biometric-adjacent portrait imagery, failing to disclose third-party transfer can lead to uninformed sharing, retention, or secondary use of personal images.

Static analysis

No suspicious patterns detected.