Eleven tools currently turn what is on your screen into something Claude Code, Cursor or Codex can act on. None is right for every job, and the honest answer depends on whether you work in a browser, need a continuous timeline, or want to pay once instead of every month.
Updated August 2026. Every price below was read off the vendor's own page on 16 August 2026 and is linked to it. Two kinds of dollar figure appear here: vendor list prices, what the software costs, and token costs, what your agent spends reading the output, always quoted at $10 per million input tokens. This comparison includes Walkie, the tool this site sells — read how this blog is written and checked first. Several of the other ten rows are the better choice for specific jobs, and this article says which.
What are you choosing between?#
Eleven tools, in four groups. Still-image capture covers plain screenshots, Vibeshots and Casso. Browser-scoped annotation covers Vibe Annotations, Markagent, Agentation and Jam. Recording-to-markdown pipelines cover Clipy and talkthrough-mcp. Narrated-review bundles cover Walkie and Cobalt Capture. They differ on price, on what they capture, and on how cheaply an agent can read the result.
That last point is the one people skip: a tool that hands your agent a full-resolution image is not free just because the app was. The category itself is covered in the pillar on visual feedback for AI coding agents; this page is the priced shortlist for the last step of the wider vibe coding workflow. Loom is deliberately omitted, because its AI reads the transcript, not the pixels; if Loom, Clipy and Walkie are the three you are weighing, the head-to-head on Walkie, Clipy and Loom compares them line by line.
What does each tool cost today?#
Free for six of the eleven. Vibeshots is $6.99 once, Casso $29 once after a $1 three-day trial, Walkie $39 once, Clipy Pro $9 a month, and Jam's Team plan $14 per creator a month billed yearly. Agentation publishes no price. Every list price below was verified on the vendor's own page on 16 August 2026.
| Tool | List price, 16 Aug 2026 | What it captures | Voice narration | Best fit |
|---|---|---|---|---|
| Agentation | Not published; source-available under PolyForm Shield 1.0.0 | Element selector, class names, position, annotated region | No | React teams who want the overlay inside their own dev build |
| Casso | $1 for 3 days, then $29 once | Screenshot with numbered boxes and notes | No | Mac or Windows users pointing at numbered spots in one image |
| Clipy | $9/mo after a free plan of 15 recordings or 2 hours | Screen recording, transcript, extracted frames, click coordinates | Transcript only | Teams wanting a shareable video link and a markdown twin |
| Cobalt Capture | Free; no paid tier published | Screenshots, typed or dictated notes, source URL | Yes, by dictation | Stills with commentary, shared as a review URL |
| Jam | Free; Team $14 per creator/mo, billed yearly | Browser session, console and network telemetry | No | Bug tickets and QA handoffs, not build review |
| Markagent | Free forever, no account | Clicked element, selector, source file path, viewport, note | No | Fast browser feedback pasted straight into an agent |
| Plain screenshots | Free | One static image | No | A single state that is not changing over time |
| talkthrough-mcp | Free, MIT licence | Narrated recording, transcript segments, keyframes, OCR text | Yes | Developers who would rather self-host and extend it |
| Vibe Annotations | Free, source-available | DOM selector, React or Vue component, viewport, screenshot | No | Localhost-only, browser-only bug reports |
| Vibeshots | $6.99 once | Screenshot capture with on-device secret redaction | No | Fast, cheap capture — not review |
| Walkie | $39 once | Recording plus voice narration, as an intent-selected frame bundle | Yes | Whole-Mac review workflows, bought once |
Sources, each read on 16 August 2026: Agentation's repository, Casso, Clipy, Cobalt Capture, Jam's pricing page, Markagent, the talkthrough-mcp repository, Vibe Annotations and Vibeshots.
Three corrections against earlier versions of this page: Clipy's free plan is not a fixed-length trial but the first 15 recordings or 2 hours, with no expiry; Cobalt Capture's price is no longer unpublished, since its capture-and-share workflow is free with no paid tier listed; and a traction statistic previously quoted for talkthrough-mcp could not be re-verified, so it is gone.
What does the agent pay to read the result?#
Tokens, and the amount depends on whether the tool sends text or an image. One 1080p screenshot costs 2,691 tokens on the high-resolution tier — Claude 4.7 and later models, which cap any single image at 4,784 visual tokens. On the standard tier the same image is downscaled to 1456x819 and costs 1,560. Both figures are Anthropic's, not measured here.
That split matters when you compare Casso against Markagent. Casso hands the agent an annotated image, so it costs image tokens every time it is sent. Markagent, Agentation and Vibe Annotations hand over text — a selector and a note — for a fraction of that. Neither is wrong; they are priced differently in a place the pricing page does not show you.
| Figure | Value | Type |
|---|---|---|
| Clipy Pro over one year | $108, at $9 per month | list price |
| Casso, after the trial | $29, charged once | list price |
| Walkie, total, ever | $39 | list price |
| One 1080p screenshot, high-resolution tier | 2,691 tokens | calculated from Anthropic's formula |
| Same screenshot, standard tier, downscaled to 1456x819 | 1,560 tokens | calculated |
| High-resolution cap on any single image | 4,784 tokens | Anthropic's published limit |
| A 30-second review frame-dumped at 1 fps | 80,730 tokens, or $0.81 at $10 per million input tokens | calculated |
| A complete Walkie bundle | about 3,500 tokens, or $0.04 at $10 per million input tokens | measured typical |
The formula is ceil(width / 28) x ceil(height / 28), capped at 4,784 on the high-resolution tier, published in Anthropic's vision documentation with the 1920x1080 row that gives 2,691. Run your own numbers in the screenshot token calculator.
Which tool is cheapest?#
Six cost nothing: plain screenshots, Vibe Annotations, Markagent, talkthrough-mcp, Cobalt Capture's free workflow and Jam's free plan. Among tools that charge, Vibeshots is cheapest at $6.99 once, then Casso at $29 and Walkie at $39. Clipy Pro reaches $108 in a year.
- Free, indefinitely — plain screenshots, Vibe Annotations, Markagent, talkthrough-mcp, Cobalt Capture's free workflow, and Jam's free plan.
- Free, then $9/month — Clipy, free for your first 15 recordings or 2 hours, then $108 a year.
- $6.99 once — Vibeshots, the cheapest paid option.
- $29 once — Casso, after a $1 three-day trial, on Mac or Windows.
- $39 once — Walkie, which also has a free trial of 7 recordings and 10 screenshots, no card required.
- $14 per creator/month — Jam's Team plan, billed yearly, aimed at teams rather than solo builders.
- Not published — Agentation, which lists no price and ships under PolyForm Shield 1.0.0.
If price alone decided this, the free rows win every time. Price does not decide it: "free" and "does the job you need" are different claims.
Which tool is best for browser work?#
Vibe Annotations, Markagent or Agentation, all of them free to start. Vibe Annotations runs a local MCP server and sends DOM selectors, component names and viewport data straight to Claude Code or Cursor. Markagent copies a structured markdown prompt to your clipboard. Agentation installs into your own React build as a dev dependency. None records voice.
Pick between them on plumbing. If you want the agent to pull annotations itself over MCP, Vibe Annotations is the fit: a Chromium extension plus a local server, scoped to localhost. If you would rather press a shortcut and paste, Markagent works on any page. If the overlay should ship inside your own React dev build, Agentation is the npm-installable one — check its PolyForm Shield terms before adopting it at work. For all three, Walkie's whole-Mac reach is capability you would pay for and not use.
Which tool is best for a whole-Mac workflow?#
Walkie, because it is not confined to a browser tab — it captures voice narration and pointer intent across any application on the Mac and compiles them into a REVIEW.md bundle of about 3,500 tokens. That reach is also its limit: Walkie is macOS only. Casso is the desktop alternative that also runs on Windows.
The honest framing is narrow. Walkie is built for reviewing something outside a browser tab across a sequence rather than a single frame. Casso covers the same desktop territory for one still image with numbered callouts, on two platforms. Clipy is the closer comparison for anyone who also wants a shareable video link.
When is a plain screenshot enough?#
When the problem is a single broken state — a wrong colour, a misaligned element, text cut off. A screenshot costs nothing to make and 2,691 tokens for the agent to read on the high-resolution tier, which caps images at 4,784. It stops being enough the moment the bug involves motion or timing.
If that one image needs three things pointed at, Casso's numbered boxes do the pointing. Narration earns its cost only when the bug is about change: an animation that stutters, a state that only breaks after three clicks.
How should you choose?#
Match the tool to what is broken, not to price or brand. One still image needs a screenshot. Browser-only work needs Vibe Annotations, Markagent or Agentation, all free to start. A shareable video link with a markdown twin needs Clipy. A review spanning more than a browser tab, bought once, needs Walkie or Casso.
- One broken state, nothing moving → a plain screenshot. Free, and already sufficient.
- Browser or localhost only → Vibe Annotations for MCP, Markagent for clipboard, Agentation for an in-app overlay.
- A still image with three things to point at, on Mac or Windows → Casso, $29 once.
- A link a teammate can watch, plus text the agent can read → Clipy, at $9 a month after the free allowance.
- Stills with dictated commentary, shared as a review URL → Cobalt Capture, free today.
- A browser bug ticket with console and network data → Jam, free or $14 per creator a month.
- Your review spans more than a browser tab, and you would rather pay once → Walkie, at $39.
None of these is a permanent choice. The cheapest useful test is to run your next real review through whichever row matches the work you are doing this week.
