ClawHub
Back to skill

Security audit

高思导引专题模块

Security checks for vulnerabilities and agentic risk

Overview

This skill is a coherent Chinese-language study-plan template with disclosed PDF export tooling, and I found no hidden persistence, exfiltration, or destructive behavior.

Install only if you want a Chinese 高思导引 study-plan generator. Expect it to create/copy HTML/PDF files, launch a local Chrome/Edge/Chromium instance for PDF export, and optionally run a localhost-only static server for viewing; avoid pointing it at untrusted HTML or unintended output locations.

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
  • Behavioral ASTexec() Call, eval() Call, Dynamic Import
  • MCP Least PrivilegeUnderdeclared Capability, Wildcard Permission, Missing Permission Declaration
  • 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 (8)

Tp4

High
Category
MCP Tool Poisoning
Confidence
99% confidence
Finding

The declared purpose is a study-plan generator, but the skill operationally instructs the agent to copy files, invoke a Python script, detect local browsers, print to PDF, and start a local HTTP server. This mismatch is dangerous because reviewers and policy layers may approve it as benign educational planning while it actually performs broader local execution and filesystem actions that can be abused, especially if user-controlled paths or files are passed through.

Content

No source excerpt is available for this finding.

Undeclared Tool Scope

Medium
Category
MCP Least Privilege
Confidence
83% confidence
Finding

The skill references shell-based execution (python scripts/build_pdf.py ..., python -m http.server) and browser-based PDF generation, but it does not declare any tool scope or allowed-tools constraints. That creates an authorization gap: an agent may infer it is permitted to use shell/environment capabilities broadly, increasing the chance of unintended command execution or filesystem interaction beyond the narrow planning use case.

Content

No source excerpt is available for this finding.

Description-Behavior Mismatch

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The manifest says the skill builds plans following a 7-module knowledge tree, but the template explicitly fixes the knowledge tree as '七/八模块' and then implements eight sections, including a separate 行程 module in addition to 应用题 and 组合数学. This is a semantic mismatch in the delivered plan artifact rather than a minor formatting detail.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
92% confidence
Finding

The document explicitly sets the page language to zh-CN and presents the title/content only in Chinese, with no indication that users may choose another language or that the template is restricted to a China-specific audience. Under the policy rules, a forced language/locale without opt-in is a natural-language policy concern.

Content

No source excerpt is available for this finding.

Intent-Code Divergence

Medium
Category
Not specified by scanner
Confidence
99% confidence
Finding

The note states the source route is based on '七大专题', but the actual table immediately below enumerates eight modules, with separate rows for 应用题, 行程, and 组合数学. This is an active contradiction between explanatory text and the artifact’s real structure.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
94% confidence
Finding

The entire skill document is written in Chinese and presents the content as the authoritative knowledge-tree reference without any indication that language choice is optional. Under the policy, forcing a specific language without user opt-in is a natural-language locale violation unless clearly documented as region-specific or justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
93% confidence
Finding

The entire workflow is written as Chinese-only operational guidance, including required output labels and examples, with no indication that the user may choose another language or that the skill is intentionally restricted to a Chinese-only regional context. This creates a natural-language policy concern because it implicitly mandates a specific language/locale without opt-in.

Content

No source excerpt is available for this finding.

subprocess module call

Medium
Category
Dangerous Code Execution
Confidence
70% confidence
Finding

subprocess module calls execute external commands. Without careful input validation, this enables command injection.

Content

Scanner excerpt · scripts/build_pdf.py (reported line 68)May include surrounding context.

python
]
    print(f"Rendering with: {browser}")
    try:
        subprocess.run(cmd, check=True, capture_output=True, text=True)
    except subprocess.CalledProcessError as e:
        print("ERROR: browser exited with code", e.returncode, file=sys.stderr)
        print(e.stderr, file=sys.stderr)