Back to skill

Security audit

teacher-agent

Security checks for vulnerabilities and agentic risk

Overview

ClawScan rates this skill suspicious because scanners report cleanup instructions that can forcibly stop unrelated Python processes and interrupt other work.

Consider the reported risk of unrelated Python work being stopped before installing. Skillspector inspected only 1 of 3 components fully and reports partial coverage, including limits in static analysis; the evidence is incomplete, so the absence of other findings is not assurance that no other risks exist.

Vulnerability Patterns
  • Insecure Skill Coding PracticesFinds exploitable flaws such as hardcoded secrets or command injection
  • 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
Findings (1)

T09 · Insecure Skill Coding Practices

Warning
Location
SKILL.md:576
Finding

Renderer cleanup forcibly terminates unrelated Python processes

Content
View full analysis

Impact

Concurrent Python services, notebooks, training jobs, or other workloads can be abruptly terminated, interrupting work and losing unsaved in-memory state.

Vulnerability Patterns
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Output HandlingUnvalidated Output Injection, Cross-Context Output, Unbounded Output
  • 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 (3)

Cleanup forcibly stops all Python processes without warning about unrelated workloads.

High
Category
Output Handling
Confidence
99% confidence
Finding

This identifies the same unsafe cleanup operation as SDI-2, rather than a separate underlying vulnerability. The skill presents broad, forced Python termination as routine cleanup without warning that unrelated workloads can be affected, increasing the chance of accidental service disruption and loss of in-memory state.

Content

No source excerpt is available for this finding.

Diagram cleanup forcibly terminates all Python processes, not just the skill's HTTP server.

High
Category
Excessive Agency
Confidence
99% confidence
Finding

The cleanup command selects Python processes by name rather than identifying the HTTP server started by this skill, forcibly terminating unrelated accessible Python processes. Routine diagram cleanup could therefore interrupt notebooks, training jobs, services, or other agents and lose unsaved work; silently suppressing errors also obscures cleanup failures.

Content

No source excerpt is available for this finding.

The student model defaults to Traditional Chinese (zh-TW) without user opt-in.

Medium
Category
Output Handling
Confidence
94% confidence
Finding

The preferences specify language: "zh-TW", and L112 instructs the agent to create new student models with these defaults. This establishes a specific output locale without asking the user or documenting a region-specific justification.

Content

No source excerpt is available for this finding.