A thought is a claim.
Write the thought out and see what it actually claims: a thinking aid built on cognitive behavioural ideas, not therapy.
What it does to the sentence, and what it will not do
It is a thinking tool that uses well-known ideas from cognitive behavioural therapy — the same ones in the self-help workbooks. It is not therapy, it is not a therapist, it cannot know you, it does not diagnose anything, and it is no substitute for talking to a person.
Claim decomposition runs in your browser, free, before anything is charged. The thought is split into its separate claims and each one is marked as an event, a reading of what something meant, a prediction, or a judgement of worth — because those four are settled by different things, and running them together is what makes a thought feel unarguable.
Eleven named wording patterns are then marked: all-or-nothing terms, another person's thoughts asserted as fact, a forecast written in the grammar of a fact, one event turned into a permanent self-label, filtering, shoulds, catastrophising, personalising, feelings treated as evidence, comparison, and overgeneralising. Every one of them is a description of the sentence rather than of the person, and every one is anchored to a quoted span of your own text, so you can see exactly what produced the mark.
A safety check runs before anything is spent. Where it fires it routes to real crisis and domestic-abuse services instead of an analysis, and it matches on the shape of a statement rather than on a list of alarming words, because the phrasings that matter most rarely contain them.
The paid step is the reconciliation. Every reframe must concede what is accurate in the original thought before disputing what is not — a reassurance that your problem is not real is not a reframe. The answer is then checked automatically for consolation posing as reasoning: universal claims about what everyone thought, unhedged promises about the future, and minimising. A second automatic check looks for clinical framing detected by construction rather than by vocabulary, so an answer cannot sound like an assessment while avoiding every clinical word.
Reconciliation between the free browser scan and the paid answer is shown too: anything the scan marked and the answer skipped is listed openly rather than quietly dropped. Template detection reports repeated wording within one answer, and across your own saved answers.
Nothing is stored unless you tick a box that resets to off every time you open the page, and one button deletes everything permanently. Whatever you do keep exports as Markdown, JSON or plain text.