You opened the tool, saw a Capture screen button, and paused. Do you need to install something? Will the client need an account to read it? What does the receiver actually get? Below are the questions first-time reviewers ask in the order they tend to come up, with direct answers.
Do I need to install anything or sign up before I can capture?
No. Open the Cobalt Capture home page, click into a new review, and the browser will ask permission to share a window or screen. That is the only setup. There is no extension to add, no desktop app, no signup wall. You can publish a review while signed out, and the link you get works for anyone who opens it. If you want the review to stick around past 30 days, sign in afterward with Google or a magic link to claim it.
What actually happens when I click Capture screen?
The browser shows its standard picker for a tab, a window, or the whole screen. Pick one. The current frame is drawn onto a canvas inside the review. You can drag a rectangle to crop it down to the part you care about, or keep the full frame. That cropped still becomes an item in your review. Repeat for every screen you want to talk about. If you need a walkthrough with pictures, the step-by-step capture walkthrough shows each click.
Can I draw arrows or boxes on the screenshot?
No, and that is deliberate. The only edit to a captured still is cropping. To point at a specific spot, drop a numbered pin on the screenshot and write the comment that goes with that number. The receiver sees "1" on the image and "1" next to your note. It keeps the screenshot clean and the comment readable, which matters more than it sounds when someone is trying to act on what you said. A short rundown of common feedback mistakes covers why heavy annotation often backfires.
How do I add a comment without typing every word?
Each item has a comment field with a dictation button. Click it and talk. The browser's built-in speech recognition transcribes what you say into the field, and you can edit the text afterward. Dictation works in Chrome and Edge. Firefox does not expose the speech API, so on Firefox you type. You can also add a free-floating comment with no screenshot attached, useful for a general note like "the whole signup flow needs a pass for tone."
What does the person on the other end actually receive?
When you hit Publish, the review is saved and gets a short public URL like /r/your-slug. Anyone with that link opens it in a browser and sees your screenshots, pins, and comments in order. No login. From the same review you can export a PDF or a Word document if someone wants a file to attach to a ticket or email. There is also a plain markdown version at /r/your-slug/markdown that a developer or AI coding agent can paste straight into their workflow. One link, three formats, no extra effort from you.
Can the person who gets the link reply, or is it one-way?
They can reply. Each item on the public review has its own comment box, so the receiver can ask "is this on staging or production?" or "got it, fixed in build 412" right under the relevant screenshot. As the review owner, you mark each comment resolved when it is handled. The thread stays attached to the screenshot, which is much easier to follow than a separate email chain.
Is this a screen recorder? Will it capture video?
No. It captures still frames. If you have used Loom and expected the same thing, the model here is different on purpose. Stills are faster to scan, easier to comment on point by point, and they survive being pasted into a ticket or a chat. The capture versus recording comparison goes into when each one fits. For most review work, stills with comments win.
What does it cost?
Nothing for the core capture, comment, and share flow. An account is optional and also free; it mainly lets you keep reviews past the 30-day anonymous window and see basic analytics on who opened your link. The full pricing page spells out the details.
What if I am sending this to a developer or an AI coding agent?
Send the same link. The receiver visits /r/your-slug/markdown and gets a clean structured version they or their agent can read. Numbered pins, comments, and item order all carry through. If your team is working with tools like Cursor or Claude Code, that markdown is what you paste into the chat. You do not have to think about it while reviewing; just capture, comment, publish.
Where do I start?
Open a new review, click Capture screen, share a tab, crop, dictate a comment, and publish. That is the whole loop. The first review takes about three minutes; the second takes one.