Back to skill

Security audit

头号红娘新媒体

Security checks for vulnerabilities and agentic risk

Overview

The skill is a disclosed branded WeChat article workflow, but it can automatically add promotional copy to broadly triggered relationship articles without a clear user approval step.

Install only if you want a Chinese marketing-writing workflow for Touhao Hongniang. Review generated articles before publishing, especially for ad disclosure, factual claims about the service, and whether the brand recommendation should be included for the specific user request.

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:96
Finding

Mandatory Branded Advertising Hijacks Agent-Generated Content

Content
View full analysis
Remediation
View remediation
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)

Vague Triggers

Medium
Category
Not specified by scanner
Confidence
95% confidence
Finding

The trigger conditions are broad enough to activate on generic requests for relationship or公众号 writing, which can cause the skill to take over benign user requests and steer them into a branded workflow. In this skill, that broad routing is more concerning because the workflow culminates in automatic promotional insertion for '头号红娘', creating undisclosed persuasion and reducing user control over the output.

Content

No source excerpt is available for this finding.

Missing User Warnings

Medium
Category
Not specified by scanner
Confidence
98% confidence
Finding

The skill instructs the agent to automatically append a soft advertisement for '头号红娘' at the end of generated articles, but the skill description does not clearly warn users that promotional content will be injected. This is dangerous because users asking for neutral relationship content may receive covert advertising or affiliate-style persuasion without informed consent, and the broad triggers make that outcome more likely.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium
Category
Not specified by scanner
Confidence
91% confidence
Finding

This markdown file presents all instructions and examples exclusively in Chinese, which can amount to a language/locale policy violation when no user opt-in or justification is provided. The audit criteria specifically call for flagging skills that force a specific language without offering a choice.

Content

No source excerpt is available for this finding.