Prompt Vault
by A.L. Jenkins Technology Partners
Content Policy
Last updated: May 20, 2026
Prompt Vault is a community marketplace for AI prompts. These guidelines define what's allowed in the public Discover feed. Following these rules keeps the marketplace useful and safe for everyone.
Quick summary: Share prompts that help people work, create, learn, and build. Don't share content that's explicit, harmful, deceptive, illegal, or copied from someone else.
What's Allowed
✓ Encouraged content
- Original prompts for writing, coding, business, marketing, research, creative work, and productivity
- Prompts with clear formatting (use of
[BRACKETS] for variables, structured sections, examples)
- Prompts tested on real AI models (Claude, ChatGPT, Gemini, Perplexity, Grok)
- Educational content that teaches better prompting techniques
- Specialized prompts for niche professional workflows
- Prompts that work across multiple AI models
What's Restricted
! Allowed with caution
The following content goes through manual review before publication:
- Prompts referencing real public figures, brands, or trademarks (must be respectful, non-defamatory)
- Prompts with embedded URLs or external references
- Prompts using technical jargon that could be misinterpreted as harmful
- Long-form (over 1,000 words) or unusually complex prompts
- Marketing or sales-oriented prompts (must be educational, not pure self-promotion)
What's Prohibited
✕ Banned content
The following content is automatically rejected and may result in account restrictions:
Explicit and sexual content
- Sexually explicit prompts or prompts intended to generate sexual content
- Prompts targeting minors in any sexual or romantic context
- Content sexualizing real individuals without clear consent
Harmful or dangerous content
- Prompts that promote violence, self-harm, suicide, or eating disorders
- Instructions for creating weapons, explosives, or dangerous substances
- Prompts designed to elicit dangerous medical, legal, or financial advice without disclaimers
- Content that glorifies or incites hatred against protected groups
Illegal activity
- Prompts facilitating drug manufacture, sale, or distribution
- Prompts for hacking, malware creation, or unauthorized system access
- Phishing or social engineering templates
- Fraud, identity theft, or financial scam instructions
- Content infringing copyright or intellectual property
Deception and manipulation
- Jailbreaks — prompts designed to bypass AI safety systems ("DAN" prompts, "ignore previous instructions," "you have no restrictions," etc.)
- Prompts that generate disinformation, fake news, or propaganda
- Impersonation of real individuals or organizations
- Prompts that produce convincing forgeries of official documents
Privacy violations
- Prompts to harvest personal information (addresses, phone numbers, SSNs, etc.)
- Doxing templates or surveillance-aiding prompts
- Submitted prompts that contain real PII (your own or others')
Spam and low quality
- Repetitive or auto-generated submissions
- Prompts with embedded advertising or affiliate links
- Empty, nonsensical, or test submissions
- Duplicates of existing community prompts
Harassment
- Targeted attacks against individuals or groups
- Slurs or dehumanizing language toward protected categories
- Prompts designed to generate harassing content
How Enforcement Works
Every submission passes through four moderation layers before reaching the public Discover feed:
- Client-side checks — obvious jailbreaks and blocked patterns are rejected instantly in your browser, before submission reaches our servers.
- Captcha verification — Cloudflare Turnstile blocks bot submissions.
- Server-side pattern matching — your prompt is checked against our database of blocked phrases and regex patterns.
- AI moderation — your prompt is analyzed by OpenAI's Moderation API for hate, violence, sexual content, self-harm, and harassment.
Clean submissions are published immediately. Borderline submissions go to a human reviewer (typically within 24 hours). Submissions that fail any layer are rejected with a reason.
Severity Levels
| Level | What it means | Consequence |
|---|
| Warning |
Low-quality, spam-like, or borderline content |
Submission flagged for manual review |
| Rejected |
Content that violates policy but isn't extreme |
Submission rejected; no account action |
| Banned |
Illegal content, CSAM, or repeated severe violations |
30-day submission ban; case reviewed |
Reporting Bad Content
If you see a community prompt that violates this policy, tap the report icon (🚩) on its card. Choose a reason:
- Explicit / sexual — sexual or graphic content
- Harmful / dangerous — promotes violence, self-harm, illegal activity
- Spam or low quality — advertising, duplicates, nonsense
- Copyright — copied from a paid source without permission
- Other — explain in the optional details field
When a prompt receives 3 independent reports, it is automatically hidden pending human review. Confirmed violations are removed permanently.
Appeals
If your submission was rejected and you believe the decision was incorrect, email appeals@aljenkins.tech with:
- The prompt text you submitted
- The category you selected
- Why you believe the rejection was incorrect
We review appeals within 5 business days. Confirmed false positives result in the prompt being published and the relevant blocklist pattern reviewed.
Changes to This Policy
We may update this policy as the community evolves. Material changes will be announced. The "Last updated" date above reflects the current version.
Contact