Red Flag: Cringe Chat Analyzer is a red flag checker for a real chat. Share a WhatsApp export or upload screenshots from any app, and the AI names the red flags it finds, each with context from the chat and how serious it is (low, medium or high), plus the pattern behind them and what to do next. It runs on iPhone; one chat costs $1.99.
Most red flag tests are quizzes about you: how you feel, what you'd forgive. A chat is better evidence. It shows who reaches out, who goes quiet, what gets promised and what happens next, all with timestamps.
What counts as a red flag in a chat
- Hot and cold: days of attention, then silence, then back like nothing happened.
- Breadcrumbing: just enough contact to keep you hoping, never a plan.
- One-way questions: you ask about their day; they never ask about yours.
- The reply-time cliff: answers that used to take minutes now take a day.
- Only late at night: warm at 1 a.m., distant at 1 p.m.
- Guilt and pressure: making you responsible for their mood, or pushing past a no.
The longer list, with what each one tends to mean, is in 10 red flags in texting.
How the checker works
- Give it the chat. Export a WhatsApp chat (Without Media) and share it to the app, or upload up to 15 screenshots of a chat from Instagram, Tinder, Hinge, Bumble, iMessage, Snapchat or any other app.
- Read the flags. Each red flag has a name, what in the chat shows it, and a severity: low, medium or high.
- See what's behind them. The pattern you two fall into, each person's attachment style, the one line that says the most, and what to do next, with a suggested reply.
Next to the flags you get the numbers, counted from a WhatsApp export: who texts more, who starts, how fast each of you replies. A flag like "slow to reply" is easier to trust when the average is right there.
Read the severity, not the count
Every read lists a few flags, because almost every chat has some friction worth naming. A low one is usually a habit: dry replies on a busy week, a double text. A high one is what to take seriously. If everything comes back low, that's a good sign, not a quiet one.
Can AI spot manipulation or gaslighting in texts?
It can point to patterns in what was written, such as guilt-tripping, denying something said earlier in the same chat, hot and cold timing, promises that never turn into plans. Across months of messages those are hard to see from the inside and easy to see on paper.
What it can't see: tone of voice, what happened in person or on calls, or what someone meant. Treat the result as a careful second opinion on the chat, not a verdict on the person.
Red flag checker vs pasting the chat into ChatGPT
A general chatbot can talk about red flags in a short exchange you paste in. With a real chat history it won't count in plain chat: asked for reply times on a three-month export, two current OpenAI models handed back a Python script to run instead. Very long chats may not fit in one message at all. The checker counts first, then reads. A side-by-side of the apps that do this is in the comparison of AI chat analyzer apps.
If the chat shows something serious
If messages threaten you, control where you go or who you see, or make you feel unsafe, that's beyond what any app should judge. Talk to someone you trust, and contact local emergency services or a domestic-abuse helpline in your country.