Security audit

文档

Security checks for vulnerabilities and agentic risk

Overview

This skill presents itself as secure encrypted document collaboration, but its only artifact is broad documentation with unsupported security claims and read/write/command authority.

Review this skill carefully before installing, especially for confidential documents. Do not rely on its encryption, privacy, collaboration, or permission claims unless the publisher supplies an auditable implementation and clear key/access-control documentation. If used at all, restrict it to non-sensitive files and require explicit approval before file writes or command execution.

SkillSpector

Vulnerability Patterns

Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
Behavioral ASTexec() Call, eval() Call, Dynamic Import
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)

Intent-Code Divergence

High

Confidence: 96% confidence
Finding: The skill prominently claims end-to-end encryption and privacy guarantees, but the file contains only marketing/documentation text and no concrete implementation, protocol description, key management design, or verifiable encryption workflow. Users may trust the skill with sensitive documents under a false security assumption, causing plaintext disclosure or unsafe handling of confidential data.

Intent-Code Divergence

Medium

Confidence: 88% confidence
Finding: The FAQ asserts real-time collaboration, permissions, and access-control capabilities without describing any actual enforcement mechanism. Security-sensitive claims about authorization and collaborative access can mislead users into assuming protections exist when they may not, increasing risk of unauthorized access or data mishandling.

Vague Triggers

Medium

Confidence: 80% confidence
Finding: The invocation description is broad and ambiguous, covering file processing, conversion, extraction, encryption-related handling, and automation without clear boundaries. In a skill that declares read/write/exec tools, vague scope increases the chance that an agent will overreach into risky file or command operations that the user did not intend.

Missing User Warnings

Medium

Confidence: 91% confidence
Finding: The skill advertises file handling and command execution, but the documentation does not present a prominent user-facing warning that using the skill may write files or run commands. In the context of enabled read/write/exec tools, missing consent and safety guidance can lead to unexpected state changes, destructive actions, or unsafe execution behavior.

Static analysis

No suspicious patterns detected.