# RateFilter

> RateFilter lets people define personal Red Lines (Filters) and analyzes available business reviews for evidence. The answer is PASS, WARNING, or FAIL. It is not a court, health inspector, lawyer or emergency service. AI output can be wrong and must be reviewed with the listed quotes.

Official site: https://ratefilter.top

## Public pages
- [Home](https://ratefilter.top/)
- [Search](https://ratefilter.top/search)
- [Public Filter boards](https://ratefilter.top/boards)
- [Privacy](https://ratefilter.top/privacy)
- [Terms and limits](https://ratefilter.top/legal)
- [Contact](https://ratefilter.top/contact)
- [FAQ](https://ratefilter.top/faq)

## FAQ by language
- [zh](https://ratefilter.top/faq/zh)
- [en](https://ratefilter.top/faq/en)
- [de](https://ratefilter.top/faq/de)
- [es](https://ratefilter.top/faq/es)
- [it](https://ratefilter.top/faq/it)
- [ko](https://ratefilter.top/faq/ko)
- [ja](https://ratefilter.top/faq/ja)
- [pt](https://ratefilter.top/faq/pt)
- [fr](https://ratefilter.top/faq/fr)

## English FAQ
### What is RateFilter?
RateFilter is not a conventional review site. You define your own Red Lines (a Filter). We analyze available reviews for evidence and return PASS, WARNING, or FAIL. Star averages do not decide for you.

### How do I create a Filter? Do I need AI?
Two paths. Structured checkboxes save without a model. Natural language lets the model generate your own filter points — not the checklist — then the model checks reviews and lists violations. Without an API key, natural language falls back to a weaker local split.

### What do PASS / WARNING / FAIL mean?
FAIL means credible evidence of a hard Red Line violation. WARNING means a preference issue or low-confidence evidence. PASS means no relevant violation was found in the reviews analyzed. Absence of evidence is not proof that a problem does not exist.

### How do public Filters, likes and boards work?
You can make a Filter public and free. Others can like, save, or copy a share link. Boards switch by likes, saves, uses, industry, language, and time window. Admins can hide a Filter without the owner noticing, or delete it permanently.

### How does language work? Can I only use same-language reviews?
The language menu needs no login. We remember your last choice, otherwise the browser language. If you enable “only evaluate reviews in this language”, search ignores other languages. Filters can also have a language tag for boards.

### What is the difference between suspend and delete?
Suspend blocks login and re-registration with the same email; data is kept. Delete removes the account and associated Filters; the email may register again. Username, avatar and password can be changed in settings. Do not share accounts.

### What data do you process? What does AI see?
We process email, username, password hash, optional avatar, locale, Filters, searches, support messages, and a session cookie. Natural-language compile and review evaluation may send filter points and review text to configured Gemini. Do not send passwords, payment data or IDs in support. See the privacy page.

### How do I contact support?
Sign in, open account settings, and message support there. Guests cannot send support mail. FAQ and language need no account. If there is immediate danger, contact local emergency services — do not wait on in-app chat.

### Where do search results come from? Are they reliable?
Businesses and reviews come from available public sources and a local catalog. Limited coverage is labelled. AI can misread reviews; open the evidence and quotes. RateFilter is not a certification, health inspection, or legal opinion.

### What can administrators do?
Admins can search accounts to suspend or delete, search Filters to hide or delete, reply to support, and configure Gemini keys and the model matrix. Hidden Filters are invisible to others without notifying the owner. Admins cannot sign in as you or change your password except via reset.