Security audit

baoyu-infographic

Security checks across malware telemetry and agentic risk

Overview

This skill is a disclosed infographic generator that writes local output files and can send generated prompts to OpenRouter for image creation.

Install only if you are comfortable with the skill saving source-derived files in the workspace and, when image generation is used, sending prompt content to OpenRouter or another image service. Review and redact confidential, personal, credential, or proprietary content before using it.

SkillSpector

By NVIDIA

Vulnerability Patterns

Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access

Findings (5)

Lp3

Medium

Category: MCP Least Privilege
Confidence: 92% confidence
Finding: The skill instructs the agent to read local reference files, read or save user-provided source content, access environment variables indirectly via a bundled script requiring OPENROUTER_API_KEY, and perform network-backed image generation, yet no explicit permissions are declared. This creates a transparency and policy gap: the platform or user may not realize the skill can access files, environment-backed secrets, and external services, increasing the chance of unintended data exposure when sensitive source material is processed.

Vague Triggers

Medium

Confidence: 85% confidence
Finding: The trigger description is broad enough to match common requests like 'visual summary' or 'poster,' which can cause the skill to activate in contexts where the user did not intend external image generation or extensive file handling. Over-broad activation increases the chance that unrelated or sensitive content is funneled into this workflow and potentially written to disk or sent to a third-party model.

Vague Triggers

Medium

Confidence: 88% confidence
Finding: The activation guidance tells the agent when to trigger but not when to avoid triggering, despite the workflow involving file reads, output persistence, and possible network transmission to an external image model. In ambiguous cases, this can over-collect content and route sensitive documents through a higher-risk pipeline than necessary.

Missing User Warnings

Low

Confidence: 87% confidence
Finding: The framework explicitly instructs the agent to save analysis results to `analysis.md` without indicating that the write is user-visible, consented to, or constrained to a safe workspace pattern. In an agent environment, undisclosed file writes can create side effects, overwrite existing files, or persist sensitive source-derived content unexpectedly, even if the intent here appears to be ordinary workflow guidance rather than abuse.

Missing User Warnings

Medium

Confidence: 92% confidence
Finding: The workflow explicitly instructs users to copy generated prompt content into third-party image generation services, but it provides no warning about privacy, confidentiality, or downstream data retention risks. If the source article or document contains sensitive, proprietary, or personal information, this step can cause unintended disclosure to external providers and loss of control over the data.

VirusTotal

65/65 vendors flagged this skill as clean.

View on VirusTotal

Static analysis

No suspicious patterns detected.