Back to skill

Security audit

图表工具专业版

Security checks for vulnerabilities and agentic risk

Overview

This diagram-generation skill is mostly purpose-aligned, but it asks for broad execution and writing authority while leaving activation, automatic versioning, and outbound callback/API behavior under-scoped.

Install only if you intend to let the agent create diagram files and run local rendering tools. Before use, narrow activation to explicit diagram/export requests, disable or confirm auto-commit behavior, and require approval for callback URLs, public PlantUML rendering, npm-based CLI execution, and any writes outside a chosen diagram output directory.

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

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The documentation presents contradictory security and dependency claims: it says no extra API keys are needed, yet later warns about API key leakage via environment variables. This inconsistency can cause operators to misconfigure secrets handling, incorrectly assume no sensitive credentials are involved, and deploy the skill without appropriate secret-management controls.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill claims batch generation is purely local and does not involve external API calls, but elsewhere advertises API integration and network-communication security guidance. These conflicting statements obscure the actual network boundary, which may lead users to permit execution in sensitive environments under the false assumption that no outbound communication or external service interaction can occur.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The trigger condition is overly broad ('project management, task planning, progress tracking, team collaboration'), while the skill has read/exec/write capabilities. Broad activation criteria increase the chance the agent invokes a high-privilege skill for ordinary collaboration requests, causing unnecessary file writes, command execution, or other side effects without clear user intent.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The skill describes file writing, command execution, and an optional callback URL, but does not clearly disclose the concrete user impact or require explicit consent before side-effecting actions. In context, this is more dangerous because the skill is designed to render diagrams via shell commands and can write artifacts or initiate network callbacks, creating real risks of unwanted local changes or outbound requests.

Content

No source excerpt is available for this finding.