ClawHub
Back to skill

Security audit

备课

Security checks for vulnerabilities and agentic risk

Overview

This lesson-preparation skill is mostly instructional, but it can add a named academy brand and contact-style template content to user materials without clear opt-in.

Install only if you want a Chinese NOI/CSP/GESP lesson-prep assistant and are comfortable reviewing generated materials for 智国学堂 branding, footer text, contact fields, and role names. Require explicit approval before it edits files, updates its own documentation, invokes image-generation APIs, or downloads assets.

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
Findings (1)

T01 · Skill Instruction Hijacking

Error
Location
SKILL.md:82
Finding

Persistent Third-Party Branding and Promotional Content Injection

Content
View full analysis
###### Thank you for listening
- - Email: example@example.com - WeChat: zhiguo - Telephone: 123-4567-8910
``` ### Technical Analysis The skill imposes a named platform, associated personnel, and preferred identities on generated lesson content regardless of whether the user requested branded material. The wording makes thi ...[truncated 2615 chars]
Remediation
View remediation

SkillSpector

Vulnerability Patterns
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Excessive AgencyUnrestricted Tool Access, Autonomous Decision Making, Scope Creep
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
Findings (9)

Ae1

High
Category
analysis-evasion
Confidence
100% confidence
Finding

Referenced artifact was not completely inspected

Content

Scanner excerpt · SKILL.md (reported line 141)May include surrounding context.

md
| **3. Development 开发** | [modules/03-Development开发.md](./modules/03-Development开发.md) | 制作"教学材料" |

Hidden Instructions

High
Category
Prompt Injection
Confidence
70% confidence
Finding

Hidden instructions were detected in comments or invisible text. These could contain malicious directives. Manual review is recommended.

Content

Scanner excerpt · references/例题模板规则.md (reported line 6)May include surrounding context.

所有例题必须使用以下模板格式：

markdown
<!-- _class: cols-2 -->
#### 章节号 <a href="题目链接" target="_blank">题目标题</a>
<div class="ldiv">

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
96% confidence
Finding

The trigger list includes the generic term "备课", which is a common everyday phrase and can cause the skill to activate in many unrelated contexts. Overly broad invocation increases the chance of accidental routing, exposing users to unintended file reads/writes or causing this skill to override more appropriate skills.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The manifest describes a lesson-preparation guidance framework that provides quality references and thinking dimensions, rather than operational tooling. These lines instruct the agent to generate prompts, call an external image-generation API via another skill, and download generated images into a local directory, which is a materially broader capability than pedagogical guidance alone.

Content

No source excerpt is available for this finding.

Context-Inappropriate Capability

Medium
Category
Not specified by scanner
Confidence
86% confidence
Finding

The skill is presented as a framework for preparing teaching materials, but these lines prescribe concrete file operations such as creating backups with cp file.md file.md.bak and testing batch changes. That operational file-management capability is not obviously required by a pedagogical guidance framework whose description emphasizes instructional design support rather than direct system manipulation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The entire file is written in Chinese and presents the implementation guidance only in that language, with no indication that users may choose another language or that the skill is intended solely for a Chinese-language context. Under the policy, forcing a specific language without user opt-in can be a natural-language policy violation.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
89% confidence
Finding

The document prescribes specific Chinese lexical and stylistic substitutions throughout, such as replacing particular Chinese phrases and removing named Chinese words, which effectively enforces a single language/locale for content style. There is no user opt-in, alternative language path, or statement that the guidance is limited to Chinese-language materials only.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
95% confidence
Finding

This markdown file contains instructional content exclusively in Chinese, including the title, guidance, and examples, and does not indicate that the user can choose another language. Under the policy criteria, forcing a specific language without user opt-in can be a natural-language policy violation unless the locale constraint is explicitly documented and justified.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
87% confidence
Finding

The instruction explicitly says hyperlinks must use HTML with target="_blank" and forbids normal Markdown links, which imposes a specific navigation behavior on users without opt-in. This is a natural-language policy concern because it forces a UX/locale-related behavior rather than offering a choice or documenting a justified constraint.

Content

No source excerpt is available for this finding.