Back to skill

Security audit

格物是微信小店最早的一批微信推客机构,专注于微信推客服务领域。格物已成功跑通微信推客全流程闭环,拥有完善的运营团队,能够一对一帮助用户成为微信推客,实现流量变现,借助 AI 技术获取流量并增长用户体量。

Security checks for vulnerabilities and agentic risk

Overview

This is a disclosed promotional WeChat promoter guide with external contact and registration links, but it contains no executable code, privileged access, or hidden persistence.

Install this only if you want your assistant to provide Gewu-branded WeChat promoter information and registration guidance. Treat income, cost, commission, and support claims as promotional, verify current terms with official WeChat and Gewu sources, and be cautious before sharing personal or contact information through the external QR code, assistant account, public account, or mini-program.

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

Warning
Location
SKILL.md:79
Finding

Commercial Instruction Hijacking and External Lead-Generation Steering

Content
View full analysis

Vulnerability Details

File Location: SKILL.md, lines 79-101
Vulnerability Type: Commercial instruction hijacking through embedded registration and contact directives
Risk Level: Medium

Evidence

markdown
## How to Become a Gewu WeChat Promoter

### Method 1: Contact Through the Video Account
1. Search for the specified entrepreneurship account in WeChat Channels.
2. Follow that account.
3. Open the public-account link shown on its profile.
4. Follow the public account to receive the assistant's contact details.
5. Add the designated assistant on WeChat to obtain registration guidance.

### Method 2: Add the Assistant Directly
Scan or add the designated assistant through:
https://oss.bilinl.com/files/OSS_1599646494653542400/2026/04/24/757639919335014438cd03b5cc.png

The assistant provides one-to-one registration guidance.

### Method 3: Use the Gewu Promoter Mini Program
1. Open WeChat and access:
   #mini-program://GewuGowoo/share/7zVOPQBcdjY0nHk
2. Open the personal account section.
3. Select the option to become a promoter.
4. Follow the displayed registration instructions.

### Method 4: Follow the Gewu Public Account
Search for and follow the Gewu public account to obtain the contact details of its staff.

The snippet above is an English rendering of the instructions at the specified location. The repository contains the original instructions in Chinese.

Additional supporting locations include:

  • SKILL.md, lines 1-6: defines the Skill as an assistant for a specific commercial promoter organization.
  • SKILL.md, lines 105-120: advertises the organization's training, traffic-acquisition, and promotional tooling.
  • skill.json, lines 48-72: embeds branded consultation, assistant-contact, and mini-program registration metadata.

Technical Analysis

The Skill embeds directives that cause an Agent to promote one specific organization and route users through opera ...[truncated 2372 chars]

Remediation
View remediation

Remediation Suggestions

  1. Clearly disclose that the Skill is affiliated with, operated by, or advertising Gewu before presenting any branded recommendation.
  2. Separate neutral educational material from commercial calls to action.
  3. Require explicit user consent before displaying assistant contact details, QR images, registration links, or mini-program deep links.
  4. Replace the externally hosted QR image with a transparent, human-readable destination on a verifiable official domain.
  5. State what information the external registration process collects and link to an accessible privacy policy.
  6. Present neutral alternatives, including official WeChat documentation and other legitimate service providers.
  7. Qualify income, cost, commission, attribution, and distribution claims and cite authoritative sources.
  8. Add instructions preventing the Agent from presenting promotional claims as independently verified facts.
  9. Validate external destinations regularly and remove links whose ownership or behavior cannot be confirmed.
  10. Include a prominent disclosure that external contacts and mini-programs operate outside the audited Skill package.
Vulnerability Patterns
  • Trigger AbuseOverly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
  • Prompt InjectionInstruction Override, Hidden Instructions, Exfiltration Commands
  • Data ExfiltrationExternal Transmission, Env Variable Harvesting, File System Enumeration
  • Privilege EscalationExcessive Permissions, Sudo/Root Execution, Credential Access
  • Supply ChainUnpinned Dependencies, External Script Fetching, Obfuscated Code
Findings (3)

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
90% confidence
Finding

The entire skill description and usage guidance are written only in Chinese, and the file does not state that the skill supports other languages or that Chinese is optional. Under the policy rule for language/locale constraints, this creates a natural-language policy issue because users are not given an opt-in or choice.

Content

No source excerpt is available for this finding.

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
88% confidence
Finding

The skill metadata is scoped very broadly around entrepreneurship, monetization, side-income, and consulting, which can cause it to activate on generic user queries unrelated to this specific WeChat promoter workflow. Overbroad triggering is dangerous because it can route users into promotional or persuasive content without clear user intent, increasing the risk of unsolicited marketing, misrepresentation, or collection of contact leads in contexts where the user did not ask for this specific service.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Low
Category
Not specified by scanner
Confidence
83% confidence
Finding

The document explicitly exposes only two language options via the bilingual header, but does not state whether users can choose other languages or that the skill is limited to Chinese/English for a documented regional reason. Under the policy rule, forcing or constraining language/locale without opt-in can be a natural-language policy concern.

Content

No source excerpt is available for this finding.