Everything you might ask before installing.

Clarishot is a browser extension that turns AI conversations into polished, publish-ready screenshots and PDFs. One click on any message in ChatGPT, Claude, or Gemini gives you a pixel-perfect capture of the whole message — not just what fits on your screen — cropped to the conversation, with none of the surrounding app chrome.

ChatGPT (chatgpt.com), Claude (claude.ai), and Gemini (gemini.google.com). Clarishot’s capture engine is built specifically around how each of these apps renders conversations — including their quirks with long chats, code blocks, and tables.

Chrome and Edge (any Chromium browser that installs from the Chrome Web Store), version 120 or newer, on desktop. It doesn’t run on mobile browsers — Chrome for iOS/Android doesn’t support extensions like this.

It’s free. A paid tier may come later for advanced features, but the core capture experience is free to use.

No. There’s nothing to sign up for, no login, and nothing to configure. Install it, open a conversation, click the camera.

Hover over any message and click the camera button that appears next to the platform’s own Copy button. The full message — even one that’s ten screens tall — is captured top to bottom and copied to your clipboard. A checkmark on the camera confirms it’s ready to paste.

Two ways:

• On the page: ⌘-click (Mac) or Ctrl-click (Windows/Linux) the camera on each message you want. Shift-click selects a whole range. Then hit Capture in the card that appears.

• From the toolbar: click the Clarishot icon and choose “Select messages.”

Non-adjacent selections get a tasteful “··· N messages omitted ···” divider, so the result reads honestly.

Yes — click the Clarishot toolbar icon and choose “Capture entire chat.” Clarishot loads the full history first (even the parts the platform hasn’t rendered yet), then captures every message. Short chats land on your clipboard as an image; long ones are saved as a multi-page PDF automatically.

Yes — Option-click (Mac) or Alt-click (Windows/Linux) the camera on either the prompt or the response, and Clarishot captures the pair as one image. Collapsed prompts are expanded automatically so nothing is cut off.

They get their own camera buttons. Code blocks (with syntax highlighting), tables (including ones wider than your screen), math equations, and Mermaid diagrams can each be captured individually — perfectly cropped, at full size, regardless of how the platform clips them on screen.

Captures land on your clipboard as an image whenever they fit within what browsers can reliably encode. Very large captures — long multi-message selections or entire conversations — are saved as a multi-page PDF instead. Clarishot tells you up front: the button says “Capture” or “Save as PDF” before you click, so there are no surprises.

Images go straight to your clipboard — paste them anywhere. PDFs go to your Downloads folder. Nothing is uploaded, stored, or sent anywhere.

Yes. Captures reproduce the conversation exactly as rendered — dark theme, light theme, custom fonts, syntax colors, all of it. That’s the point: pixel-perfect means your pixels.

No. No analytics, no telemetry, no accounts, no servers. Clarishot has no backend at all — every capture is composed locally in your browser, and the output goes only to your clipboard or your Downloads folder. Your conversations never leave your machine.

Short answer: Chrome’s screenshot API demands it. The captureVisibleTab API that photographs the page requires that broad permission — Chrome offers no narrower version of it that works with in-page buttons.

What matters in practice: Clarishot’s code only ever runs on chatgpt.com, claude.ai, and gemini.google.com (you can verify this in the extension’s manifest — the content scripts are restricted to exactly those three sites), and it talks to no Clarishot servers — there are none. The broad-sounding permission is an API requirement, not a data grab.

It only reads. Clarishot observes the conversation to place its camera buttons and photographs the page during capture — it never alters your messages, never touches your account, and never modifies what the platforms send or receive. During a capture the page scrolls (that’s how full-length messages get photographed) and then returns exactly where you were.

No. Clarishot’s buttons only respond to real user input — scripts on a page can’t simulate clicks to trigger captures — and capture output goes only to your clipboard or downloads.

Refresh the conversation tab. Browser extensions can’t attach to tabs that were already open before the extension was installed or updated — a refresh fixes it. The toolbar popup will tell you exactly this if it detects the situation.

Captures photograph the visible page, so the tab needs to stay visible while the capture runs. Switch tabs or hide the window mid-capture and Clarishot cancels cleanly rather than producing a broken image — nothing partial ever lands on your clipboard. Press Escape if you want to cancel one on purpose.

Chrome rate-limits screenshots to two per second, and a long conversation can take dozens of them. Clarishot shows live progress and an up-front time estimate for entire-chat captures — and you can cancel anytime with Escape.

Pinch/accessibility zoom distorts what the screenshot API sees, so Clarishot declines to produce a blurry, cropped capture rather than handing you a bad one. Reset zoom to 100% and capture again.

Capturing a message while the AI is still writing it would give you a cut-off answer. Clarishot waits until the response is complete — try again the moment it finishes.

The clipboard write happens at the very end of a capture and requires the page to still be focused. If you switched to another app at exactly the wrong moment, the camera icon reverts (no checkmark) — that’s Clarishot telling you the copy didn’t land. Click capture again and let it finish before switching away.

Not yet. Chromium-only for now.

No — and that’s deliberate. Clarishot is a publishing tool for AI conversations, not a general screenshot utility. The narrow focus is what makes the captures this good.

The AI platforms ship UI changes constantly, and extensions like this one live downstream of that. We run automated checks against the live platforms to catch breakage early, and fixes ship as extension updates — if something looks off, refresh the tab first (it solves most of it), then tell us at support.

A full-page tool photographs the whole page — sidebar, header, composer included — and you crop out the conversation afterward. Clarishot works at the conversation level: one message, a prompt with its answer, a hand-picked set, a single code block or table, or the entire chat. Already cropped, nothing to clean up.

Long chats matter too. ChatGPT and Gemini unload messages that scroll far off-screen, so scroll-and-stitch tools can save blank or missing sections without noticing. Clarishot loads the full history first, checks its output for gaps, and fails visibly rather than saving a broken image.

Exporters read the text out of your chat and rebuild it in their own document template — that’s why they offer font sizes and switchable themes. What you get is a document that resembles your conversation, not the conversation itself. Clarishot captures what the platform actually rendered: real fonts, real colors, real syntax highlighting.

Privacy is the other difference. The popular exporters we’ve checked generate PDFs on their own servers and cap free PDF exports at a few a day. Clarishot composes everything locally — no servers, no account, no quota.

Messages and conversations are photographed with Chrome’s screenshot API — the pixels come straight off your screen, stitched into one tall image. Nothing is re-typeset or run through a template.

Wide content is the one exception. A table wider than your window has pixels the browser never painted, and a screenshot can’t include pixels that don’t exist — so Clarishot renders those elements at full size from the live page instead. Wherever a photograph is possible, you get a photograph.