You need to flag six things on a staging page and send them to a developer by lunch. Snagit can do that. So can a browser tab with nothing installed. The right pick depends on whether the work is mostly annotating images for yourself, or producing a review someone else has to read and act on.
Snagit is a paid desktop capture app from TechSmith. It is genuinely good at what it does: rich image markup, scrolling capture, templated step-by-step images, a local library of past captures. A browser-based capture and share flow like Cobalt Capture is a different shape of tool. It runs in a tab, captures a still frame, lets you talk or type a comment against it, and publishes a public review link plus a PDF, Word doc, or markdown export. No install. No signup required. Free.
Below is where each one earns its keep, and where it gets in the way.
What you actually produce at the end
Snagit produces image files. You capture, mark up with arrows and boxes and text callouts, and save a PNG or paste it into Slack, Jira, or an email. The annotations live on the image itself. If the receiver wants to comment back, they reply in whatever tool you pasted into.
A browser capture and share flow produces a review. Each screenshot is an item with its own comment, optionally a numbered pin pointing at a spot, and the whole thing publishes to a short URL like /r/abc123. The receiver opens the link, reads it in order, and can post a reply on any individual item. You can mark each reply resolved. The same review is available as a PDF, a Word document, or plain markdown, which is the format an AI coding agent can read without any conversion.
If your output is mostly one polished image to drop in a doc, Snagit wins. If your output is a list of issues someone has to triage, a browser-based alternative to Snagit saves the receiver real time.
Image annotation depth
Snagit is the clear winner on the image itself. Arrows, callouts, blur for sensitive data, stamps, shapes, step numbers baked into the pixels, scrolling captures of a long page, simplify mode that turns a UI into abstract shapes. If you publish tutorials or marketing imagery, those tools matter.
Cobalt Capture deliberately does less to the image. The only image edit is cropping the captured still. Instead of drawing on the screenshot, you add a numbered pin that points at the spot and put the explanation in a comment next to it, typed or dictated through the browser's built-in speech recognition. The comment carries the meaning, not annotation layered on pixels. That is faster to produce and easier to read on a phone, but it is not a replacement for a polished annotated image.
Install, license, and who can use it
Snagit is a paid app with a per-seat license. Installing it on a work machine usually means a request to IT. A freelancer reviewing a client site, a PM on a borrowed laptop, or a stakeholder you want feedback from once cannot reasonably be asked to install and license a desktop app. If your reviewers face locked-down machines, the no-extension capture workflow matters more than annotation depth.
A browser-based flow has none of that friction. Open a tab, click Capture screen, share the window, crop, comment, publish. Anonymous reviews are kept for 30 days, and signing in later claims them permanently. Pricing is free, and an account is optional.
Speed for a one-off review
For a single annotated screenshot, Snagit is fast once it is already open. For a review of eight things across three pages that needs to land in someone's inbox as one link, the browser flow is faster end to end. You skip launching the app, saving files, naming them, uploading them somewhere, and writing the surrounding context in an email. Capture, talk, publish, paste link. Five minutes start to finish.
This is the same logic that applies to sending client feedback or a staging site walkthrough: the receiver should be able to act from the artifact alone.
Handing work to a developer or an AI agent
A Snagit PNG dropped in Jira asks the developer to look at the image, read your callouts, and translate them into a fix. That works. It is what teams have always done.
If part of the work goes to an AI coding agent, image annotations are a dead end. The agent reads text. A review published as markdown with screenshot links is something a tool like Cursor or Claude Code can ingest directly. See how feedback reaches an agent for the mechanics. For human-only teams this does not matter; for hybrid teams it matters a lot.
Which one suits you
Stay with Snagit if you produce documentation, training material, or marketing imagery where the annotated image is the deliverable; if scrolling capture of long pages is core to your work; or if you already own a license and the friction is gone for you and your reviewers.
Switch to a browser capture and share flow if your output is a list of issues another person reads and acts on; if your reviewers will not install software; if you want one link that holds context, threaded replies, and exports to PDF, Word, or markdown; or if any part of the work goes to an AI coding agent. Start a review in a new tab and see whether your next round of feedback needs more than that.