Back to skill

Security audit

fashion-lookbook-generator

Security checks for vulnerabilities and agentic risk

Overview

The skill mostly matches its fashion-image purpose, but it includes a risky shortcut that can copy the newest clipboard screenshot without the user explicitly choosing it.

Install only if you are comfortable with the agent using web/ImageGen workflows and avoid relying on the clipboard shortcut; provide or approve the exact reference image yourself before generation, especially if your clipboard screenshots may contain private information.

Vulnerability Patterns
  • Unauthorized Access and Privilege EscalationObtains permissions beyond the task's legitimate needs
  • 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
Findings (1)

T05 · Unauthorized Access and Privilege Escalation

Warning
Location
SKILL.md:65
Finding

Automatic Selection of the Newest Clipboard Image Without Explicit User Approval

Content
View full analysis
Remediation
View remediation
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 (7)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger list contains broad, everyday phrases that could cause the skill to activate in contexts the user did not intend. Because the skill performs web fetching, local file access guidance, and image-generation workflows, unintended invocation can lead to unnecessary data access or actions beyond user expectations.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The default model value is set to “东亚年轻女性,” imposing a specific ethnicity, age, and gender presentation without indicating user choice or opt-in. This is a natural-language locale/demographic constraint that the policy says not to force unless the user is given a clear option or the limitation is justified.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The skill instructs reading from a local path outside the workspace (~/.workbuddy/clipboard-images/) to locate the latest clipboard screenshot. That is a local file access pattern unrelated to the declared working directory boundary, and it can expose sensitive user images or other content without a clear, informed consent step.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill describes reading clipboard images from a local directory but does not provide a clear warning or obtain explicit consent before inspecting local files. Clipboard-derived screenshots often contain sensitive personal or business information, so silent or assumed access creates a privacy and data-exposure risk.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill includes broad browser automation and shell/network instructions (agent-browser, curl, grep, md5 checks) that go beyond simple image-generation behavior and materially expand what the agent may do on the user's behalf. In context, fetching a Pinterest image is related to the feature, but the unconstrained command patterns increase attack surface for arbitrary web access, command execution, and misuse if the skill is triggered on untrusted input.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The reusable prompt template explicitly specifies “a young East Asian woman,” which bakes a fixed ethnicity and demographic into generated outputs. Because the file does not frame this as an optional example or provide user choice at that point, it represents a natural-language policy issue around forced demographic constraints.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Low
Category
Not specified by scanner
Confidence
80% confidence
Finding

Referencing OAuth API materials in a local skills directory expands the skill's effective capability surface beyond its stated image-generation role and may encourage use of local credentials or integration assets not scoped to this skill. While less severe than direct command execution, it creates a pathway to credential misuse or unauthorized external API access.

Content

No source excerpt is available for this finding.