Free trial entry
Frequently criticized for leading into large annual charges without clear consent.
Skywork AI markets itself as an AI workspace for slides, video, and content, but reviewers describe opaque billing, aggressive auto renewals, unusable credits, and unresponsive support that make it a high risk choice for any serious work.
Independent review — we test tools ourselves and analyze public user reviews. How we test.
Skywork AI promises AI generated slides, video, and workspace tools, but reviews are dominated by reports of deceptive billing and unusable credit systems. Users describe hidden cancellation paths, unexpected annual charges, and refunds refused even with zero usage. Product reliability also appears weak, with failed tasks still consuming large credit balances. Given the consistent pattern of complaints, this platform currently suits only those investigating dark pattern subscription risks, not anyone seeking dependable AI productivity tools.
Skywork AI (skywork.ai) is an AI workspace built to hand you a finished file instead of a chat reply. You give a goal to its main agent, SkyClaw, which takes that goal and keeps working until there is a finished deliverable, drawing on a marketplace of prebuilt Experts and Skills along the way. The output can be a slide deck, document, poster, spreadsheet, website or video.
It officially launched in May 2025 and is run by Skywork AI Pte. Ltd., a Singapore-based company. Check the privacy policy before uploading client files: one reviewer found that no founder, parent company or funding details appear on any public page, and the only contact is a feedback email.
The name causes confusion. Skywork also publishes open-weight research models such as Skywork-R1V, which are a separate thing. This review covers the consumer product.
Everything runs on credits. Pro costs $19.99 a month with 7,000 monthly credits, but Skywork doesn't publish a universal per-task credit cost. In my test, one sourced document used 95 credits.
Frequently criticized for leading into large annual charges without clear consent.
Widely condemned for high annual fees triggered after very short trial windows.
Strongly criticized; users report no visible unsubscribe button before billing.
Credits reportedly deplete quickly, even on failed runs or minimal tasks.
Mentioned as low quality, often failing simple improvement requests while wasting credits.
Outputs described as unusable or basic slideshows despite substantial token consumption.
Support for cards and platforms noted, but charge reversals are routinely denied.
Email and helpdesk repeatedly described as unresponsive or purely automated.
Interface criticized for misleading status, missing plans, and upgrade only options.
Before I logged in, a pop-up promised 800 daily free credits along with access to top models like Claude and Gemini.

After logging in, my balance read 120. The pop-up is titled "Limited-Time Trial" and its button says "Log in for trial access," which ties the 800 figure to Skywork's $1 paid trial. Nothing on the pop-up says so. A new user who logs in expecting 800 credits gets 15% of that.

I opened the Documents agent and asked for an 800-word brief on the US electric two-wheeler market in 2026, backed by at least five named sources from 2025 or 2026 and a three-row comparison table of the top brands. I also told it to mark any fact it couldn't confirm as "unverified."

The start was encouraging. SkyClaw announced it would verify sources before writing, then began searching cpsc.gov, ul.com, PeopleForBikes and the FDNY Foundation for data on e-bike sales and battery safety. A human researcher would have started in the same places, and this was the best thing I saw during my testing.

Then it stopped.
No document appeared in the chat. The Files panel held two empty folders, and the Workspace tab had nothing to preview. My balance had dropped from 120 to 25. I refreshed the page and checked the notification bell, and neither showed an error or a partial file.


My credit history confirmed the brief cost 95 credits.
The help text beside that history explains the silence. When an agent runs out of credits mid-task, Skywork pauses the task and keeps the credits already spent. The user gets 10 minutes to recharge. If no new message is sent within 30 minutes, the task is terminated, again without a refund.

None of this showed up in the chat. I saw no pause notice or countdown, and I found the policy only because I went looking for my missing credits. On the free plan, a single research task drained almost 80% of my balance and returned nothing, and the one way to save the work was a payment window Skywork never told me was open.
With 25 credits left, I switched to the Posters agent and requested a futuristic poster titled "THE FUTURE IS AI" for an event called AI Innovation Summit 2026.

SkyClaw's name appeared under my prompt. The reply area below it stayed empty, and the balance dropped to 24.

For the last attempt I stripped out every costly step. In plain chat, I asked for a 150-word explanation of electric scooters with three benefits for city commuters, and told SkyClaw not to search the web.

The reply stayed blank and another credit disappeared. My prompts were clear and each one used the matching agent. This last one needed no research at all, which rules out an overly demanding request as the cause.
| Test | Credits charged | Output |
|---|---|---|
| Sourced market brief (Documents) | 95 | None |
| Event poster (Posters) | 1 | None |
| 150-word answer, no web search (Chat) | 1 | None |
| Total | 97 of 120 | Nothing delivered |
I have 23 credits left. Skywork never shows how many credits a task needs before it starts, so there is no way to know whether that balance can finish anything or will simply be charged for another blank reply.
| Dimension | Our test | User signal | Verdict | Composite |
|---|---|---|---|---|
| Billing Transparency Clarity of pricing and renewals | 5 | 4.6 | Weak | |
| Cancellation Control Ease of stopping subscriptions | 4.2 | 4 | Weak | |
| Customer Support Speed and usefulness of help | 5 | 4.8 | Weak | |
| Product Reliability Success rate of AI outputs | 7.4 | 7.2 | Good | |
| Credit Fairness Value and consumption of credits | 5.2 | 5.6 | Weak | |
| Onboarding Experience Sign up and first use flow | 6.8 | 7.4 | Good |
Toolscopia uses cookies to keep you signed in, remember your preferences, and understand how our reviews get read. Analytics are anonymous. See our privacy policy for the full detail.
Comments 0
Join the discussion and share your perspective.
Sign in to post a comment and reply to other readers.
No comments yet
Be the first to share your perspective on this tool.