Back to skill

Security audit

Edupath

Security checks for vulnerabilities and agentic risk

Overview

This is a Chinese postgraduate-planning report skill with disclosed live search/data-source requirements and no executable code, persistence, or hidden install behavior.

Install this skill if you want a China-focused kaoyan planning assistant that performs many live Baidu and education-data lookups. Avoid giving unnecessary personal details, because search queries may be sent to external services. If your agent loads GEMINI.md or CLAUDE.md globally, review whether you want their general coding/file-editing guidance applied outside this skill.

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
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • MCP Tool PoisoningHidden Instructions, Unicode Deception, Parameter Description Injection
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (5)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
87% confidence
Finding

The file defines the agent as a specialized consultant for a Chinese education company and frames its outputs around Chinese postgraduate planning context from the outset. While region-specific specialization can be acceptable, the instructions do not explicitly state that this locale/language constraint is intentional for a China-only use case or offer any user opt-in/choice, which can create a language/locale policy concern.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill’s documented behavior significantly exceeds its stated purpose of generating postgraduate planning advice by mandating extensive real-time searches and operational workflows. This expansion increases the attack surface, creates unnecessary tool-use pressure, and can be abused to trigger excessive external calls or operational side effects unrelated to the user’s request.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
97% confidence
Finding

Including file modification, replace/write guidance, and git recovery procedures in an advisory skill is dangerous because it grants or normalizes workspace mutation without a legitimate business need. If the agent follows these instructions in a broader tool-enabled environment, it could alter local files, damage workspace integrity, or be steered into destructive changes under the guise of routine execution.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The role and output instructions are written to require a specific Chinese-speaking persona and Chinese colloquialisms, and the file does not provide an opt-in or alternative language option. This can violate language/locale policy when users are not given a choice.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The skill mandates real-time external searches driven by user-provided academic context, but it does not warn users that their input may be transmitted to Baidu or other third-party sources. This creates a privacy and transparency issue: a user's education background, interests, and planning details could be disclosed externally without informed consent, especially because the workflow requires repeated, granular searches.

Content

No source excerpt is available for this finding.