Back to skill

Security audit

几何情绪窗口海报

Security checks for vulnerabilities and agentic risk

Overview

The skill provides a coherent photo-to-poster workflow with ordinary installation and photo-privacy considerations.

Use an image provider you trust with your photos. The optional global installation command skips prompts and uses changing dependencies; prefer reviewed versions or the documented manual installation.

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
Vulnerability Patterns
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
Findings (4)

Description-Behavior Mismatch

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding

The supplied chunk is an installation command, not an implementation of the described poster-generation workflow. Its observable behavior changes the installed skill environment globally. Although the installed skill's name matches the advertised purpose, its contents are not supplied, so its actual image-design capabilities cannot be verified.

Content

No source excerpt is available for this finding.

MCP server referenced without pinned version: 'npx skills'.

Medium
Category
MCP Rug Pull
Confidence
90% confidence
Finding

The installation command uses npx skills without a pinned version, allowing a changing npm package release to execute with the installing user's permissions; a compromised release could therefore execute malicious code. The detector incorrectly describes this CLI as an MCP server, but the underlying dependency-pinning concern is valid. This is a conventional, manually invoked installation example rather than automatic skill execution, which reduces exposure, and there is no evidence of malicious intent.

Content

No source excerpt is available for this finding.

The prompt specifies Chinese poster titles without an explicit user language choice.

Medium
Category
Output Handling
Confidence
85% confidence
Finding

The instruction specifies a short Chinese title rather than letting the user choose the title language. Although the document labels itself a Chinese-title version, it does not explicitly condition that language requirement on user opt-in or provide a justification for the restriction.

Content

No source excerpt is available for this finding.

The prompt requires an English title without explicit user language opt-in.

Low
Category
Output Handling
Confidence
90% confidence
Finding

The instruction to produce “one short English title” imposes a specific output language regardless of the user's preference. Although the heading labels this an English-title version, the file provides neither an explicit language choice nor a justification for requiring English.

Content

No source excerpt is available for this finding.