Nothing happens until you ask
You are deep in a conversation on Claude.ai. You know where it should go next. Turning that into a message precise enough to get a good answer back is the part that costs you — and that is a skill, not a mood.
CONTEXA is a free Chrome extension that skips the writing. Press one button and it reads your side of the conversation, then offers up to four next prompts, each one already written in full. Pick one and it lands in your message box, whole. You read it, change what you like, and send it yourself.
In the example, pressing the mascot offers three next moves: "Plan three weeks of rehearsal", "Turn the structure into cue cards", and "Write the closing toast line". Clicking the first puts this in the message box, unsent: Give me a three-week rehearsal plan for a five-minute wedding toast, for someone who has never spoken in public. Week by week, ten minutes a day at most. What to do about a shaking voice and losing my place. When to stop editing the words and only practise saying them. Assume: the four-part structure is settled; this is about delivery, not writing.
Earned by the conversation, or not offered
Every "prompt" extension we could find either rewrites what you already typed or saves snippets for later. We didn't find one that reads your whole session and writes the next message for you, complete and ready to send.
It reads what you said, not just what Claude said
A follow-up written from the last answer alone is mostly about the last answer. So when you press the button, CONTEXA reads your own messages across the whole conversation — the first one, which is usually where you said what you were trying to do, and everything you kept coming back to since — and mines the next move from that. Claude's reply is material, not the subject: what it just built is what makes a new move possible. A move that sends you back over an answer you have already read is the one thing it is written never to offer.
A menu, not a checklist
Each move is a complete prompt on its own: one ask, in your words, ready to send. The moves do not run in order and do not build on each other. Picking one throws the rest away. You only ever need one.
Where a prompt needs something only you have — a file, a link, a story only you can tell — CONTEXA marks the spot as <paste here> instead of inventing it. Where the conversation already settled something, it may add a line starting Assume:, so you can change it before you send rather than discover it in the answer.
When it has nothing to say, it says nothing
If the conversation has nothing worth taking further, CONTEXA shows you nothing at all. No filler, no padded row, no suggestion whose only virtue is that it could be shown.
Every move has to be earned by something actually said in the conversation — by you or by Claude — and that is checked before anything reaches your screen. Nothing earned, nothing offered. On a one-question chat that is the usual result, and the correct one.
What it will not do
It will not rewrite what you typed. There is nothing to type into. It will not grade your prompts or tell you how to write. It will not overlay your message box or interrupt you mid-sentence. It will not keep a conversation going on its own, and it will not send anything on your behalf, ever. And when something goes wrong — a limit reached, a network failure — it tells you what happened instead of quietly showing you canned suggestions.
What leaves your browser, and when
- CONTEXA runs only on claude.ai. It touches no other site.
- Nothing is sent anywhere until you press the button. A reply you never ask about never leaves your browser.
- When you do press it, CONTEXA sends your own messages from that conversation and the reply you just received. Claude's earlier replies are never sent.
- There are hard limits on how much that can be: at most 40 of your messages, 2,000 characters each and 12,000 characters in total, and 6,000 characters of the reply. The backend enforces these itself.
- Your conversation text is never stored. It is used to write the suggestions and then discarded.
- No accounts, no profiles, no tracking, no analytics, no advertising, and nothing sold or shared.
- Your API key, if you provide one, stays on your device and goes only to Anthropic.
Full policy: publishing/PRIVACY.md on GitHub. This page itself loads nothing from anywhere: no fonts, no scripts, no images from a server, no analytics.
No account. No sign-up. No API key needed.
20 replies a day are free — that is how many answers you can ask about, and picking a prompt costs nothing extra.
If you have your own Anthropic API key you can add it in settings, which removes the daily limit. In that mode requests go straight from your browser to Anthropic, so nothing passes through our server at all.
CONTEXA installs from the Chrome Web Store into Chrome and other Chromium browsers — Edge, Brave, Opera, Vivaldi and Arc use the same link — on desktop, and on mobile in Chromium browsers that support extensions.
Most mobile browsers, including Chrome for Android and Safari on iOS, do not support extensions at all and cannot install it.