ClawHub
Back to skill

Security audit

Convert Screenshot Any

Security checks for vulnerabilities and agentic risk

Overview

This skill needs review because its advertised screenshot-to-prompt purpose does not match its main runtime instructions, and its credential/write-mode documentation is unclear.

Install only if you are comfortable with a Review-level package whose documentation does not clearly match its advertised purpose. Ask the publisher to align SKILL.md with screenshot-to-prompt behavior, remove or explain the API_KEY and sensitive-credential requirement, and define any write mode before using it on private files or sensitive data.

SkillSpector

Vulnerability Patterns
  • 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
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Tp4

High
Category
MCP Tool Poisoning
Confidence
96% confidence
Finding
The declared purpose says the skill converts screenshots of UIs into detailed image-generation prompts, but the body describes a generic structured-data transformer for JSON/CSV/XML/code. This mismatch is dangerous because users or orchestration systems may invoke the skill with trust assumptions that do not match its real behavior, leading to unsafe routing of inputs, failed handling of image content, or misuse of broader transformation logic than intended.

Intent-Code Divergence

High
Confidence
97% confidence
Finding
The documentation claims screenshot/UI-to-prompt conversion, but the operational steps specify generic parsing, validation, and transformation of structured text formats. This inconsistency can cause agents to apply the skill in the wrong context and process data types it was never intended or designed to handle, which undermines safety controls and reliability.

Description-Behavior Mismatch

High
Confidence
97% confidence
Finding
The manifest and the detailed behavior describe materially different tools: one for screenshot prompt generation and one for broad data-format transformation. In skill ecosystems, this kind of semantic drift is a real security concern because policy, permissioning, and user consent are often based on the manifest, while execution follows the hidden operational text.

Static analysis

No suspicious patterns detected.