ClawHub
Back to skill

Security audit

Khan Tutor

Security checks for vulnerabilities and agentic risk

Overview

This is a tutoring-only Markdown skill with no code, installs, credential access, persistence, or hidden data movement.

Install this if you want responses to lean toward Socratic tutoring and practice generation. Be aware it may activate for broad learning-help phrasing and may offer related study aids such as flashcards or quizzes, but it does not itself install or run those integrations.

A.I.G

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

SkillSpector

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

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The trigger phrases and scope are broad enough that the skill could activate for many generic education-related requests, causing unintended routing and over-application of this skill. In an agentic system, overly broad invocation can override more appropriate skills, leak contextual assumptions across tasks, or produce behavior the user did not intend, reducing reliability and potentially affecting safety-sensitive workflows if tutoring style is applied in the wrong context.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Low
Category
Not specified by scanner
Confidence
90% confidence
Finding

Lines L023-L024 establish a strict tutoring principle of not giving answers first, but later sections direct the agent to "Always show every step" in worked examples (L137-L151) and to provide a "partial solution" as Hint 4 (L161-L164). That is an internal contradiction in the documented intent about how directly the tutor should reveal answers.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.