What Clarq does, and what it doesn't.
What does Clarq do?
It rewrites the draft in your message box into a clear, specific prompt. It first works out what you're asking for (coding, debugging, writing, research, learning, business, creative, or image and video), then adds what a good answer needs: the goal, the context, constraints and the format you want back.
It also formats the prompt for the site you're on, for example tagged sections for Claude and a short search question for Perplexity. Nothing is sent until you press send.
How does it make answers more precise?
AI assistants answer the question they're given. A vague request gets a generic answer, or a guess at what you meant. Clarq states the goal and the expected output, and lists the details that matter for that kind of request, such as the exact error, your environment and what you've already tried.
Where it doesn't know a detail, it leaves a marked placeholder like [paste error here] instead of making something up, so the assistant works from your facts.
Does it stop hallucinations?
No tool can promise that, and Clarq doesn't fact-check answers. It does reduce the usual causes. Prompts never invent facts about your situation. Research prompts ask for sources and an explicit note of uncertainty or disagreement, and Perplexity prompts ask for cited sources.
That makes made-up answers less likely and easier to spot. Still check anything important.
Does it keep my conversations consistent?
Every prompt gets the same clear structure (goal, context, requirements, format), so the assistant gets consistent instructions from one message to the next.
Clarq doesn't remember your earlier messages or conversations: each enhancement works from what's in the message box. In a long chat, use it on each new request rather than once at the start.
Does it save tokens?
Usually, through fewer messages. A vague prompt often leads to a round of "Could you share the error?" or an answer you have to correct, and every extra turn makes the assistant reread the conversation so far.
The improved prompt is longer than your draft, so a single message uses a little more input. The saving comes from getting a usable answer sooner.
What is the label next to the button?
After enhancing, Clarq shows the kind of request it detected, like "Debugging". If that's wrong, pick another and Clarq rewrites your original text again for that kind of request. This uses one more enhancement.
Is it really free?
Yes. Every account gets 100 enhancements a month, reset on the 1st at 00:00 UTC, and there's nothing to buy. A failed enhancement doesn't count. If you use all 100, enhancing pauses until the next month.