other
Warning
- Location
SKILL.md:43- Finding
Generation of Unverified First-Person Earnings and Endorsement Claims
- Content
View full analysis
Vulnerability Details
File Location:
SKILL.md:43-65,SKILL.md:94-106,SKILL.md:142-150,README.md:27-57, andprompts/titles.txt:1-13
Vulnerability Type: Deceptive Content Generation
Risk Level: MediumVulnerable Content
The following English rendering preserves the complete meaning and structure of the affected example in
SKILL.md:43-65:markdown ### Output Format # Title: I used AI automation and earned 800 yuan in 7 days (complete process included) Body: Do you often hear that AI can make money but do not know where to start? I used three steps to earn 800 yuan in seven days. The method is simple: 1. Tool selection: OpenClaw + Xiaohongshu API 2. Content strategy: Focus on tool reviews + case analysis 3. Monetization channels: Brand partnerships + affiliate commissions Earnings breakdown: - Days 1-3: Learning the tools (time investment) - Day 4: Published the first post, received over 2,000 views, and earned a 50-yuan commission - Day 7: A brand contacted me and offered 500 yuan Key point: Real cases + reproducible methods If you also want to try it, comment "toolkit" and I will send you the detailed configuration. #AI-Side-Hustle #OpenClaw #Xiaohongshu-Monetization #Side-Hustle #Technology-BlogThe content-generation rules in
SKILL.md:94-106require:markdown **Title formula**: - Number + AI tool + time + earnings - Example: "Three AI tools that let me earn more than 5,000 yuan per month" - Example: "No coding required: use OpenClaw to automate a side hustle (tutorial included)" **Body structure**: 1. Introduce the reader's pain point 2. Present my results using numbers and screenshots 3. Break down the method into three steps 4. Provide important details and warnings 5. Include a call to action asking readers to comment for a toolkitThe complete affected title-template set in
prompts/titles.txt:1-13is:text # Title ...[truncated 3838 chars]- Remediation
View remediation
Remediation Suggestions
- Require all first-person earnings, audience, acceptance-rate, commission, and brand-offer claims to come from user-provided and verifiable data.
- When verified data is unavailable, explicitly label all figures as hypothetical examples, sample projections, or fictional demonstrations.
- Remove default language such as “personally tested,” “my earnings,” “real case,” “doubled my income,” and “my results” unless the user confirms that the statements are accurate.
- Replace first-person factual templates with neutral alternatives, for example:
- “Illustrative workflow for using AI in a side project.”
- “Hypothetical seven-day example; actual results vary.”
- “Potential monetization channels, not guaranteed earnings.”
- Add a generation rule prohibiting fabricated screenshots, testimonials, analytics, commissions, brand contacts, and financial records.
- Add a mandatory disclaimer that earnings are not guaranteed and depend on skill, effort, audience, platform rules, and market conditions.
- Require citations or remove unsupported optimization statistics concerning click-through, save, completion, and engagement rates.
- Add a final verification checkpoint asking the user to confirm every personal, financial, and performance claim before publication.
- Ensure calls to action do not imply that a toolkit or configuration will be delivered unless the user actually has that material and intends to provide it.