Oxygen 6.2 Review: MCP Agents, and What Classic Users Need to Know

Oxygen 6.2 is the release where Oxygen became an AI-agent-driven builder. Connect Claude Code or Codex over MCP and an agent can build pages, templates, Components, selectors, variables, dynamic data, interactions, and animations — and change your site settings while it is at it.

But if you are one of the many people running Oxygen Classic on client sites, the MCP headline is not your story. Your story is that Oxygen 6 is a different product wearing the same name, and you need a deliberate position on what happens to your existing builds.

TL;DR

• 6.2 is the MCP release. Agents build pages, templates, headers, footers, Components, selectors, CSS variables, dynamic data, interactions, animations, and site settings.
• Clients supported: Claude Code and Desktop, Codex CLI and Desktop, Cursor, OpenCode, Windsurf, Cline, Gemini CLI, and any MCP client.
• Version status: Moving fast. 6.1 stable in May, 6.1.1 on July 15, 6.2 on July 20, 6.2 Beta 2 on July 25–26. Check the releases page before quoting a version to a client.
• Oxygen 6 is a rebuild, not an update. Backwards compatibility with Classic was broken deliberately.
• No official Classic → 6 migration. JSON exports and copy-paste do not carry over. Classic addons do not run in 6.
• Classic is still supported — bug fixes, security, and compatibility updates continue, and it has its own site at classic.oxygenbuilder.com.
• Pricing: One-time lifetime licences, $129 to $199.50. Existing lifetime holders get Oxygen 6 free.
• Recommendation: New builds on 6. Stable Classic sites stay on Classic unless there is a concrete reason to rebuild.

What 6.2 actually shipped

The MCP integration is the substance of this release. Setup lives at Oxygen → Settings → Agents & MCP, with a one-click install for Agent Connector — the same companion plugin Breakdance 3.0 uses, which tells you something about how Soflyy is building this across its product line.

Once connected, the scope of what an agent can touch is broad:

The design-system line is the one that matters most in practice. An agent that reuses your existing selectors and variables instead of generating a parallel set of one-off styles is the difference between a useful collaborator and a machine that quietly wrecks your CSS architecture. This is also where the 6.2 point releases have been doing visible work — Beta 2 added support for :where() and :is() selectors in MCP and fixed a set-global-settings tool bug that could break global settings saves.

That last item is a useful reminder of where this feature is maturing: a tool that can corrupt global settings is exactly the class of bug you want found in beta rather than on a client site.

The same admin-credential caveat applies

Connecting an agent means generating a WordPress application password, and that credential carries administrator-level access. Not builder-scoped — full admin, because that is how WordPress application passwords work. An agent holding it can do anything an admin can through the REST API.

Agent Connector lets you disable the connector when idle, lock it to a specific domain, and log MCP events for audit. Use all three, and issue the credential from a dedicated user you can delete after a build rather than from your own account.

Performance work

6.2 also brings caching for generated CSS, faster page loading, and dynamic data that loads asynchronously without blocking the canvas. Beta 2 refined this further — CSS caches no longer regenerate when a valid stored cache is empty, which specifically helps pages with many Components, and the Variables Panel got more responsive. Again, these are builder-side gains. Front-end performance is a separate conversation.

Oxygen 6 is not an update. It is a rebuild.

This is the part that gets lost in release-note coverage, so let us be direct about it.

Oxygen 6 is a ground-up rewrite of the Oxygen visual builder. It shares a name with Oxygen Classic and, in Soflyy's own framing, not much else technically. The company's launch FAQ does not soften this:

"Breaking backwards compatibility with Oxygen Classic was the hardest decision we've ever made."

What that means concretely:

• Classic JSON exports do not import into 6. There is no built-in fallback, and copy-paste between the two does not work either.
• Classic addons do not run in 6. The third-party ecosystem built up around Classic over the years — and it was substantial — does not transfer. If a client site depends on a Classic-only addon, that dependency has to be solved before any move.
• There is no one-click upgrade. Moving a Classic site means recreating templates and layouts, not converting them.

Third-party tools have appeared to fill the gap — TransferForge and WPConverters both sell Classic-to-6 conversion. They can automate structural translation, and that is genuinely useful. But read their own positioning carefully: the output requires visual styling, CSS adjustment, and design refinement afterwards. Their honest framing is that a conversion is a starting point, not a finished rebuild. Budget accordingly, and pilot on one site before you sell a client on a portfolio-wide migration.

Classic is supported, and that changes the calculus

The reason none of this is a crisis is that Soflyy has not abandoned Classic. Its stated commitment is that bug fixes, security updates, and compatibility enhancements continue unabated, that Classic users keep receiving technical support, and that sites built on Classic can continue using it indefinitely. Classic has its own documentation site at classic.oxygenbuilder.com and is maintained as a live product.

That is a materially different situation from a builder being sunset. When a vendor deprecates a product, staying put becomes an accumulating security liability and rebuilding is forced eventually. When a vendor commits to long-term maintenance, staying put is a legitimate engineering decision rather than deferred debt.

So the question is not "when must I migrate." It is "does Oxygen 6 offer enough to justify the cost of a rebuild on this specific site." For most stable, working Classic sites, the answer is no.

Cases where moving does make sense:

• The site is due for a redesign anyway. If you are rebuilding the design, rebuild it in 6. The migration cost collapses to near zero because you were going to redo the layouts regardless.
• You want agent-driven workflows on that site. MCP support exists in 6, not Classic.
• The site is small. A five-page brochure site is a couple of hours in 6. A 60-template membership site is a project.
• Standardising the whole portfolio has real operational value. One builder across all client sites means one set of skills, one update process, one set of quirks to remember.

Cases where you should stay on Classic:

• Complex sites with heavy Classic addon dependencies.
• Anything where the client will not fund a rebuild and the site works fine.
• Sites you inherited and do not fully understand yet.

Pricing, and how it compares

Oxygen's licensing is its strongest competitive argument, and it is worth laying out plainly:

All tiers are one-time payments with no recurring fees, unlimited sites including client sites, and a 60-day no-questions-asked money-back guarantee. Existing lifetime licence holders get Oxygen 6 as a complimentary upgrade. Soflyy has run promotional pricing as low as $99 for an unlimited lifetime licence earlier in 2026, so it is worth checking whether an offer is live before buying at list price.

Compare that to the sibling product. Breakdance is $199.99/year for unlimited sites. Over five years that is roughly $1,000 versus Oxygen's $199.50 one-time — and both now have comparable MCP support and a shared technical lineage. For an agency planning to build on one of these for years, the lifetime model is difficult to argue against on pure economics.

Where Breakdance still earns its premium is WooCommerce. Its visual WooCommerce template builder — single product, shop and category archives, cart, checkout, My Account, order confirmation — remains the most complete of any developer-focused builder, and it has a longer production track record on its current codebase.

The verdict

Oxygen 6.2 is a strong release on a product that is still young. The MCP integration is well-designed — reusing existing selectors and variables rather than generating parallel styles is the correct architectural choice, and the breadth of what an agent can touch goes further than most competitors. Combined with lifetime pricing from $129, it is one of the better value propositions in the WordPress builder market right now.

Start new projects on Oxygen 6. The version churn is real — four releases between May and late July — so pin a version per project, test updates on staging, and do not chase every beta.

Leave working Classic sites on Classic. The rebuild cost is real, the compatibility break is absolute, and Soflyy's long-term support commitment means there is no ticking clock. Migrate opportunistically, when a redesign is already funded, not as a project of its own.

And take the admin-credential question seriously on both builders. Handing an AI agent a full-privilege WordPress credential is a new category of operational risk for most agencies, and the tooling to scope it properly does not exist in WordPress yet.

Related reading

• Breakdance 3.0: MCP-Ready AI Agents Inside Your Page Builder
• Breakdance vs Bricks Builder 2026
• Best WordPress Page Builders 2026
• WordPress Staging Site Guide
• WordPress Client Onboarding Process