绘本生成 Picture Book

• Outcome: Pass
• Latest audit: Sep 2, 2026 · 2:34 AM UTC
• Version: 1.3.11

Overview

This skill is a disclosed picture-book generator that uses a third-party dLazy CLI, remote generation APIs, local output files, and a stored API key.

Install this only if you are comfortable using dLazy's hosted service, sending prompts and any reference images to dLazy, spending dLazy credits, and storing or supplying a dLazy API key. Run it in a fresh folder so generated book files do not overwrite unrelated work.

SkillSpector

By NVIDIA

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• MCP Least Privilege: Underdeclared Capability, Wildcard Permission, Missing Permission Declaration
• MCP Tool Poisoning: Hidden Instructions, Unicode Deception, Parameter Description Injection
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration

Findings (5)

Lp3

Medium

• Category: MCP Least Privilege
• Confidence: 91% confidence
• Finding: The skill clearly instructs the agent to create and modify local files such as `book.json`, `index.html`, `images/*`, and `music/*`, but it does not declare corresponding permissions. That mismatch can cause the platform or user to underestimate the skill's filesystem access and creates a transparency and containment problem if the skill is invoked unexpectedly or in a sensitive directory.

Intent-Code Divergence

Medium

• Confidence: 93% confidence
• Finding: The documentation states the skill itself does not exceed network or filesystem access, yet the prescribed workflow explicitly sends prompts and reference images to remote endpoints and uses curl and local file writes to assemble output. This is dangerous because it misrepresents the skill's actual trust boundary, which can cause operators to approve or invoke it without understanding that local content is uploaded and artifacts are written to disk.

Vague Triggers

Medium

• Confidence: 78% confidence
• Finding: The skill description uses broad trigger phrases around picture books and storybooks without narrowing conditions, which can cause the skill to activate for generic storytelling requests. Unintended activation is risky because it may push the agent into installing tools, requesting credentials, making network calls, or writing files when the user only wanted simple text generation.

Vague Triggers

Medium

• Confidence: 84% confidence
• Finding: The trigger keyword list is ambiguous and lacks scope limits, making ordinary requests like '睡前故事' or 'storybook' eligible for activation even when no image, music, or file assembly is desired. In this skill context, accidental activation increases exposure to unnecessary external API use, credential setup, and local artifact creation.

Vague Triggers

Medium

• Confidence: 76% confidence
• Finding: The trigger phrases include broad, everyday terms like `storybook`, `illustrated story`, and `bedtime story`, which can cause accidental invocation in ordinary conversation. Unintended activation is risky here because the skill can install or run third-party tooling, authenticate to an external service, send prompts/media off-box, and write multiple local files.

VirusTotal

63/63 vendors flagged this skill as clean.

Static analysis

No suspicious patterns detected.