
Sider Code Review: Can AI Rewrite Any Website?
Sider Code gave me one of those tests where the result changed my opinion halfway through. I started with a simple goal: use plain English to change real websites without touching HTML, CSS, or browser scripts. That is exactly what Sider Code is built for. It reads the page, understands its structure, makes changes, and can save those customizations for later visits. Sider says it can hide distractions, reshape layouts, add actions, extract information, and change how a webpage works without manual coding.
I first tried it on Claude.ai. My request was visual: turn the page into a red-and-white Optizeno theme and place floating Optizeno branding around the interface. The first result was disappointing. Most of Claude stayed almost unchanged, while Sider added only a few floating Optizeno labels. It technically followed part of the instruction, but it did not feel like a real theme redesign.

I gave it a second chance with a much more detailed prompt. This time I clearly asked it to change the background, sidebar, chat box, buttons, borders, spacing, and overall color system. The result was better. The page finally had a visible red-and-white look, a larger Optizeno card, themed accents, and a more customized background.
Still, the design quality did not impress me. The layout felt uneven rather than carefully balanced, and one large “Optizeno Intelligent Workspace” element sat over the original chat area. That directly went against my instruction not to block Claude’s controls. So for pure visual redesign, Sider Code looked capable but not reliable enough to trust with a polished interface in one shot.

The Al Jazeera test changed the mood completely. I asked Sider to move Iran–US conflict news to the priority area, then split the remaining stories into two columns: War News and Normal News. I also told it not to invent stories and to work only with content already visible on the page.

This was the strongest part of my test. Sider correctly identified the Iran–US story and placed it at the top. More importantly, it separated war-related stories from normal news even when some headlines were indirect. That showed real page understanding rather than simple keyword matching. It then created the two-column layout I requested, kept the story cards readable, and produced a page that was genuinely easier to scan.
The styling was still fairly basic. I would not call the new news layout beautiful. However, it was useful, organized, and much closer to what I had described. Functional restructuring is where Sider Code performed much better than visual decoration during my testing.
Another surprise was the credit usage. My free account started with 30 credits. Sider’s help center currently says free users receive 30 basic credits per day. I ran two Claude customizations and the Al Jazeera restructuring, yet my visible balance stayed at 30 and I saw no credit deduction. Sider’s current credit rules say newer Lite, Pro, and Max plans deduct credits based on actual usage, with detailed records available through the Credits Log. So the zero-cost result during my test was welcome, but the reason was not obvious.
I also saw an in-product offer promising 1,000 credits for sharing a strong Sider Code creation on X. I did not complete that reward process, so I cannot confirm whether those credits are awarded instantly or whether extra conditions appear after sharing.
Pricing is more complicated than the free test suggests. As of September 3, 2026, Sider’s official pricing page lists Lite, Pro, and Max as its current paid plan family and says all plans include the main features, while higher tiers mainly increase credits and cloud-computer power. Current listed USD prices show Lite at $32/month or $324.99/year, Pro at $65/month or $649.99/year, and Max at $129.99/month or $1,000/year. Pricing can vary by platform or region, so checking the live checkout before paying is sensible.

The sweet spot here is repeated website cleanup and restructuring. Researchers can organize dense pages. Readers can remove clutter. Heavy browser users can build custom views without writing scripts. Sider itself also says saved customizations can return when you revisit the same site, which could make these changes useful beyond a one-time experiment.
Pros
- Understands webpage content surprisingly well.
- Good at reorganizing complex pages.
- Works with normal language prompts.
- No coding knowledge needed.
- Customizations can be saved for later visits.
- My three tests used no visible credits.
Cons
- Visual redesign quality was weak.
- The Claude layout lacked symmetry and polish.
- Sider ignored part of my placement instructions.
- A branded element blocked the original Claude chat area.
- The first Claude result barely changed the actual interface.