Community Guidelines

Design Theater is a public space for Vibes, Prompts, Skills, Dialogues, project collaboration, and the people who build them. These guidelines describe what we expect from everyone here and how we moderate.

This policy applies to Dialogues, public discussions, comments and replies, project collaboration, Vibes, Prompts, Skills, profiles, and uploaded files.

It does not create any contract or legal entitlement, does not promise absolute protection, and does not replace applicable law. For account terms see the Terms and Privacy Policy.

How to read this

Every rule below has a stable identifier like COM-01 so moderators, notifications, and the public can point to the same thing. Rules are grouped into twelve categories and sorted into three severity levels. Severity drives the default enforcement recommendation; context and intent can soften or sharpen that default.

Severity levels

Critical — Immediate safety or security risk. Exposed credentials, doxxing, credible threats, severe abuse. Default enforcement is immediate removal and, where appropriate, account suspension pending review.

Serious — Hate speech, targeted harassment, impersonation, malicious links, unauthorized sharing of private information. Default enforcement is removal, a direct warning, and escalating restrictions on repeat.

Moderate — Spam, inappropriate promotion, minor abuse, misleading submissions, repeated low-level disruption. Default enforcement is removal or hiding of the content with an explanation; patterns of repetition escalate to Serious.

Policy categories

1. Respect and interpersonal conduct

Disagreement is welcome. Contempt is not. Argue with ideas, not with people.

• CON-01 · Moderate · Personal attacks that insult, demean, or dismiss another member rather than engaging with their work.
• CON-02 · Moderate · Deliberately derailing a thread, pile-ons, or coordinated downvoting of a single creator.

2. Harassment, bullying and targeted abuse

• HAR-01 · Serious · Targeted harassment of a specific person, including repeated unwanted contact across threads, chat, or comments.
• HAR-02 · Serious · Encouraging others to harass a specific person or project.
• HAR-03 · Critical · Severe abuse, sexualised harassment, or sustained campaigns intended to drive someone off the platform.

3. Hate speech and discrimination

• COM-01 · Serious · Hate speech, slurs, or content attacking people on the basis of race, ethnicity, national origin, religion, caste, disability, gender identity, sexual orientation, age, or similar protected characteristics.
• COM-03 · Serious · Discriminatory jokes, memes, or "ironic" content that targets a protected group.

4. Threats and violence

• SAF-01 · Critical · Credible threats of violence against a person, group, or property.
• SAF-02 · Critical · Content that promotes, glorifies, or incites violence, terrorism, or self-harm.
• SAF-03 · Serious · Instructions for building weapons, exploits intended to cause harm, or similar operational uplift.

5. Privacy and doxxing

• PRI-01 · Critical · Doxxing. Publishing someone's home address, workplace, phone number, or other identifying private information without consent.
• PRI-02 · Serious · Sharing someone's private messages, photos, or recordings without consent.
• PRI-03 · Serious · Linking a pseudonymous account to a real identity against the account holder's wishes.

6. Personal, financial and health information

• PFH-01 · Serious · Posting another person's financial details, government IDs, or medical information.
• PFH-02 · Moderate · Posting your own sensitive personal information in public contexts where it is likely to be misused. We may hide the content and send you a note; this is for your protection, not punishment.

7. Credentials, API keys and confidential information

• SEC-01 · Critical · Exposed credentials. Posting API keys, access tokens, private keys, passwords, .env contents, or any secret that grants access to a system.
• SEC-02 · Critical · Posting employer-confidential or client-confidential material without authorization.
• SEC-03 · Serious · Soliciting other members' credentials, session cookies, or private tokens.

If you discover an exposed credential in your own work: rotate it immediately at the issuing provider. Deleting the post is not sufficient; third parties may have cached it.

8. Copyright, licensing and attribution

• IPR-01 · Serious · Uploading material you do not have the right to redistribute (code, images, writing, model weights, datasets).
• IPR-02 · Moderate · Omitting attribution for work you forked, remixed, or derived from. Forking is encouraged; silently reclaiming credit is not.
• IPR-03 · Moderate · Misrepresenting the license of a work (claiming MIT when the upstream is GPL, etc.).

Formal copyright complaints follow the DMCA process.

9. Spam, manipulation and unsolicited promotion

• SPAM-01 · Moderate · Spam. Repeated identical or near-identical posts, bulk tag stuffing, keyword-only replies, SEO-only comments.
• SPAM-02 · Moderate · Inappropriate promotion. Repeated unsolicited links to products, services, newsletters, affiliate codes, or paid courses outside of a Dialogue that invited them.
• SPAM-03 · Serious · Vote or follow manipulation. Operating sockpuppet accounts, coordinated upvote rings, or paid engagement.

10. Impersonation and deceptive behaviour

• IMP-01 · Serious · Impersonation. Pretending to be another member, a public figure, a company, or Design Theater staff.
• IMP-02 · Serious · Deceptive handles, display names, or avatars designed to be confused with a real account (@openai vs @0penai).
• IMP-03 · Moderate · Misleading submissions. Claiming an output came from a model, pipeline, or tool it did not come from. Honest "I used X to help with Y" framing is fine; falsified provenance is not.

11. Malicious links, files and code

• MAL-01 · Serious · Links to malware, phishing, drive-by downloads, cryptojackers, or deceptive redirect chains.
• MAL-02 · Serious · Uploading files or code intended to compromise other users' devices, accounts, or data (RATs, info-stealers, cookie exfiltration, keyloggers).
• MAL-03 · Moderate · Content that quietly abuses the platform (hidden iframes, autoplay audio traps, pop-under loops) rather than being overtly malicious.

12. Responsible AI-generated and AI-assisted content

Design Theater is built around AI-assisted work. These rules are about using it honestly.

• AI-01 · Moderate · Non-consensual synthetic likeness of a real, identifiable person in a demeaning, sexual, or defamatory context.
• AI-02 · Serious · Deepfakes or voice clones used to deceive about facts, impersonate, or defraud.
• AI-03 · Moderate · Presenting unverified model output as factual claims in Research or benchmark contexts without disclosure.
• AI-04 · Moderate · Prompt-injection payloads or jailbreaks posted with the apparent purpose of weaponizing them against other services; sharing them as research with context is welcome.

Enforcement principles

Detected flags are not confirmed violations. Our automated filters (keyword, pattern, rate) raise signals. A human reviews before a serious or critical action is taken on a user's account.

Context and intent matter. Quoting a slur to criticise it is different from using it. Discussing an exploit in a security Dialogue is different from offering it as a weapon. Sharing a redacted screenshot of a bug is different from doxxing. Moderators consider the surrounding thread, the user's history, and plausible intent.

Escalation is gradual. The default path on moderate violations is: content removed → direct note explaining the rule → repeated behaviour tightens restrictions (rate limits, posting cooldowns) → suspension → account closure. Critical violations can skip steps; nothing else should.

Warnings include an explanation. Any removal or restriction notice names the specific rule ID (for example SPAM-02) and links to this page.

Appeals are allowed. Any enforcement action can be appealed through Report / Feedback with kind=other and a reference to the rule ID. Appeals go to a different reviewer wherever possible.

No permanent ban from a detector alone. Automated systems alone do not permanently close accounts. Permanent closures require human review and are reserved for Critical violations or sustained patterns of Serious violations.

The policy is public and versioned. This page is accessible from the site footer, the sign-up flow, and every moderation notification. Material changes are called out at the top of this page when they ship.

Reporting

If you see content that violates these guidelines, use Report / Feedback and reference the rule ID if you know it. Reports can be filed signed-in or anonymously. We do not share reporter identities with the reported party.

For urgent safety concerns (credible threats, exposed credentials, doxxing), file a report with kind=abuse or kind=id_theft and mark it clearly in the subject line. Critical reports are reviewed with priority.

What we do not promise

We cannot guarantee that every rule violation is caught, that every report is resolved within a specific window, or that moderation outcomes will always feel fair. We can promise that the rules are written down, that enforcement is explained in terms of these rules, and that appeals get a second look.