How it works

1. Pick a check and paste your text

Each check has its own page. Paste a message, post or short description, up to 1,500 characters.

2. A machine answers several narrow questions

Instead of asking an AI for an opinion, each check asks an AI model built for fast, structured judgments a handful of small questions about the text, for example:

• Should I send this? Tone, regret risk, neediness, passive-aggression, clarity.
• Is this cringe? Trying too hard, humblebragging, emoji overload, forced slang, "thought leader" voice.
• Am I overreacting? How serious the trigger was, how big the reaction was, catastrophizing, assumed intent.
• Is this a red flag? Control, boundaries, respect, honesty, pressure.

3. Fixed rules turn the answers into a verdict

The answers are combined by fixed rules into a score and a verdict. The rules look at the overall picture and at the strongest signals, so one big problem isn't hidden by everything that's fine. The same answers always give the same verdict. The "Why" line shows the signals that mattered most, and the one-liner is picked from pre-written lines, not generated.

Serious situations

If a text mentions threats, violence, abuse, stalking or self-harm, there's no score and no joke. The page shows where to find help instead, and sharing is switched off.

Limits

It's a snap judgment of the words on the screen. It only hears one side, it doesn't know the people or the history, and it works best in English. The examples include cases where it gets things wrong.

What happens to your text

Your text goes from our server to the AI provider for evaluation, and the verdict comes back to you. Our application does not store it. Details are on the privacy page.