Stress-test your writing before it goes public

Before you publish an essay or article, the assistant reads it the way a hostile critic would: hunting for factual errors, weak arguments, and the lines someone would quote to dunk on you. You fix the problems in private, not in the replies.

What it does

• Reads your draft as an adversary, not a fan, from more than one angle so it catches errors a single read would miss.
• Flags factual claims that are wrong or unverifiable, and says what is actually true where it can.
• Points out weak arguments: leaps that do not follow, straw men, false equivalences, and how an opponent would exploit each one.
• Names the attack surface: the line that gets quoted out of context, the comment that picks it apart, the angle a critic would take.
• Returns a numbered list of issues rated by severity, with a recommended fix for each, then applies the fixes you approve.

The reveal

You finish a piece you are proud of and paste it in. Back comes a list: two facts that were subtly wrong, one argument with a hole a critic would drive through, and the exact sentence that would get screenshotted and mocked. You fix all three before anyone else ever sees them.

What it doesn’t do

It does not soften your argument or add hedges you did not write. It fixes errors and weak spots while keeping your voice and conviction intact. It is a pre-publish gut check, not a ghostwriter.

Configuration

• The draft: paste the full essay or article in the chat.
• How you respond: reply to the issues by number. Anything you do not mention gets the recommended fix, so you only engage with what you care about.

Triggers

On demand, run before you publish. Hand over a draft whenever you have one close to done.