Back to skill

Security audit

Learn Graph

Security checks for vulnerabilities and agentic risk

Overview

This is a simple learning-method skill that guides users through building a knowledge graph and does not request sensitive access or execute code.

Before installing, consider that this skill may be invoked for many broad learning-planning requests and may respond best in Chinese unless the agent adapts language. It appears appropriate for users who want structured learning guidance and knowledge-map style study planning.

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
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • 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 (2)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The trigger conditions are very broad ('systematically learn X', 'where to start', 'learning path', 'overview of a field') and can match many ordinary educational requests without clear boundaries. In an agentic environment, this can cause inappropriate or overly eager skill activation, steering conversations into this skill when another tool or a direct answer would be more appropriate.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

The skill is written entirely in Chinese and implicitly assumes Chinese interaction, but it does not state that it is locale-specific or instruct the agent to follow the user's language preference. This can lead to language mismatch, degraded usability, and accidental exclusion of users, especially if the skill is invoked in multilingual contexts.

Content

No source excerpt is available for this finding.