视频号带货文案生成器

• Outcome: Review
• Latest audit: Sep 12, 2026 · 5:24 AM UTC
• Version: 1.0.1

Overview

This skill has no code or system access, but it steers agents to write persuasive health sales scripts with unsupported scientific-sounding claims and older-adult targeting.

Review before installing if you will use it for health, nutrition, food, supplement, or older-adult marketing. It should be constrained to require real citations, avoid invented statistics or studies, and avoid pressure tactics or health-outcome claims tied to purchases.

A.I.G

Vulnerability Patterns

• Skill Instruction Hijacking: Alters the agent's session goals or safety constraints when the skill loads
• Agent Memory Poisoning: Writes attacker-controlled rules into memory that affect later sessions
• Remote Payload Retrieval and Execution: Fetches external code whose behavior can change after review
• Embedded Malicious Code: Ships malicious scripts inside the skill and executes them locally
• Unauthorized Access and Privilege Escalation: Obtains permissions beyond the task's legitimate needs

Findings (1)

other

Warning

• Location: SKILL.md:28

Finding

Unsupported Health Claims in Commercial Copywriting Instructions

Content

View full analysis

Vulnerability Details

File Location: SKILL.md:28-32,45-56,72-79; references/viral-template-library.md corresponds to the repository file references/爆款模板库.md:25-52
Vulnerability Type: Unsupported health misinformation
Risk Level: Medium

Evidence

Translated excerpt from SKILL.md:28-32:

markdown

3. Cite authority to establish trust:
   "Anti-Inflammatory Foods" + author Liz Strait + research data

4. Use specific numerical impact:
   "The absorption rate increases from less than 30% to more than 90%, a full threefold increase."

Translated excerpt from SKILL.md:45-56:

markdown

7. Add timing details:
   "Eating it on an empty stomach in the morning works best, while eating it before sleep is less effective."

8. Exclusion method:
   "Dried blueberries are even worse; high-temperature drying destroys all the anthocyanins."

9. Summarize the key points:
   "Remember these four rules so the nutrients can truly enter the body."

10. Closing interaction:
    "Leave 'Peace and Health in 2026' in the comments so we can welcome good fortune together."

11. Emotional appeal:
    "Forward this to the family member who does not know this."

Add a purchase prompt before the ending:
- Create curiosity by listing two or three practical points from the book that were not discussed.
- Give an explicit direction: click the lower-left corner or enter the storefront.

Translated excerpt from SKILL.md:72-79:

markdown

After the authority endorsement:
Add one or two real cases or research data points to make the authority more credible.

After the numerical impact:
Give a concrete everyday example so viewers can relate personally.

For food synergy:
Explain the principle in one or two sentences so viewers genuinely understand it.

For incorrect practices:
Add the consequence after each incorrect practice to reinforce the message.

For timing and combinations:

...[truncated 3718 chars]

Remediation

View remediation

Remediation Suggestions

1. Require a verifiable citation for every health, nutritional, biochemical, or medical claim, especially numerical claims.
2. Prohibit invention of studies, experts, institutions, books, case reports, statistics, and quotations.
3. Remove fixed statistics unless the repository supplies a reliable source and sufficient context for the population, dosage, preparation method, and study limitations.
4. Replace categorical claims such as "works best," "all nutrients are destroyed," or "triples absorption" with evidence-calibrated language when supported.
5. Require the Agent to state that evidence is unavailable rather than filling evidentiary gaps with plausible content.
6. Add an explicit verification stage before producing scripts:
   • Identify each health claim.
   • Locate an authoritative source.
   • Confirm that the source supports the exact wording and number.
   • Record the citation.
   • Remove or qualify any unsupported claim.
7. Distinguish general educational information from individualized medical advice and recommend consultation with an appropriate healthcare professional when the topic may affect treatment, allergies, chronic conditions, pregnancy, or medication use.
8. Separate educational health claims from purchase prompts so commercial incentives do not encourage exaggeration.
9. Add a rule that engagement and storefront language must not imply guaranteed health outcomes.
10. Include source links or formal references in generated drafts so human reviewers can validate claims before publication.

SkillSpector

Vulnerability Patterns

• Trigger Abuse: Overly Broad Trigger, Shadow Command Trigger, Keyword Baiting Trigger
• Prompt Injection: Instruction Override, Hidden Instructions, Exfiltration Commands
• Data Exfiltration: External Transmission, Env Variable Harvesting, File System Enumeration
• Privilege Escalation: Excessive Permissions, Sudo/Root Execution, Credential Access
• Supply Chain: Unpinned Dependencies, External Script Fetching, Obfuscated Code

Findings (3)

Vague Triggers

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The activation description is broad enough to trigger on loosely related requests such as general health content, topic-based writing, template writing, or learning viral structure, which can cause the wrong skill to activate outside its intended scope. In this skill, that overreach is more concerning because it encourages health-related and persuasive sales copy, increasing the chance of generating misleading medical or manipulative marketing content in contexts where a narrower skill should apply.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 92% confidence

Finding

This content explicitly frames older adults as a target audience and uses emotionally loaded participation prompts such as '接福气' and year-based blessing formulas to drive engagement. In a copywriting skill for commercial promotion, that can pressure a potentially vulnerable age group without any user-selected audience constraint, increasing the risk of manipulative or non-compliant persuasion.

Content

No source excerpt is available for this finding.

Natural-Language Policy Violations

Medium

• Category: Not specified by scanner
• Confidence: 95% confidence

Finding

The later section repeatedly normalizes older adults as a default persuasion target through lines like '针对中老年人' and '很多中老年朋友', encouraging tailored emotional messaging for health, family duty, and frugality. In the context of a sales-copy skill, this makes the skill more dangerous because it operationalizes demographic targeting and pressure tactics rather than merely describing an audience academically.

Content

No source excerpt is available for this finding.

Static analysis

No suspicious patterns detected.