Build on Webflow with your favorite agent

reddit logodocusign logoabm logothe new york times logoideo logo

Take this Figma file and build it exactly as designed in Webflow as a CMS template.

Develop new pages and sections, manage CMS content, update site settings, and deploy from any MCP-compatible client.

Get started
Read the docs

Build at the speed of AI

Create pages, components, and entire site sections agentically.

Build on-brand

Use the full context of your existing design system, so every output looks like it was built by your team.

Build with content

Create, update, and manage CMS content at scale — without touching the editor. Pull from external sources, bulk edit entries, and keep your content pipeline moving.

How it works

Lorem ipsum dolor sit amet

Connect your tools

Connect Webflow to your favorite AI tools like Claude or Cursor. No custom code, no middleware.

Define your rules

Use Agent Instructions to set the skills and rules that keep every agent on-brand and on-policy. Reference your design system, CMS content, voice and tone guidelines, or legal docs.

Start building

Prompt your agent to design pages, manage content, analyze performance, and more. Everything it builds respects your design system, your permissions, and your brand, automatically.

Get started
Read the docs

Webflow MCP for Enterprise

Build with control

Webflow MCP respects your existing roles and permissions, supports branching for safe iteration, and logs every AI action so you always know what happened.

Build for visibility

Manage meta titles and descriptions for AEO.

Build to convert

Ask questions about your site performance the same way you'd ask a colleague. Then let agents act on the answers.

Customer story

From brief to breakthrough: giving Frontify's product story a pulse with Rive

How OFF+BRAND built a scroll-driven, click-reactive interactive experience in Rive for Frontify.

Read the full story

Prompt library

Create pages, components, and entire site sections agentically. Use the full context of your existing design system, so every output looks like it was built by your team.

01
Create a landing page for our summer campaign using our existing component library.
02
Pull this Figma design and rebuild it in Webflow using our design system variables.
03
Add a new section to the homepage hero — same style as the existing one.
04
Update the card component across every page to use the new border radius variable.

Create, update, and manage CMS content at scale — without touching the editor. Pull from external sources, bulk edit entries, and keep your content pipeline moving.

01
Create 10 new blog post entries from these briefs and populate all the CMS fields.
02
Update the featured image and meta description on every post tagged 'product update’.
03
Archive all CMS entries that haven't been published in over a year.
04
Pull our latest case studies from Notion and create CMS entries for each one.

Ask questions about your site performance the same way you'd ask a colleague. Then let agents act on the answers.

01
How is my site performing this week compared to last week?
02
Which pages have the highest drop-off rate right now?
03
Post a weekly performance summary to our #marketing Slack channel every Monday.
04
Find underperforming pages and suggest experiments to improve them.

Webflow MCP respects your existing roles and permissions, supports branching for safe iteration, and logs every AI action so you always know what happened.

01
What changes were made to the homepage in the last 48 hours and who made them?
02
Create a branch for this redesign so I can review before anything goes live.
03
Which team members have editor access to the checkout page?
04
Why is the nav broken on mobile? Show me what changed recently.
FAQs

Webflow MCP (Model Context Protocol) connects AI tools like Claude, Cursor, and Windsurf directly to your Webflow projects, so you can design, build, and manage your site through natural language. Describe what you want, and your AI agent handles the implementation — no manual API calls, no custom authentication setup required.

It's built on Webflow's Data and Designer APIs, which means your AI agent works with your real site data, design system, and CMS content. Whether you're a developer looking to move faster or a marketer who wants to update content without touching the editor, Webflow MCP gives your AI agent the context it needs to take meaningful action on your site.

Webflow MCP supports design, content, governance, and performance workflows — all through natural language prompts. On the design side, you can create layouts, work with components, apply styles, manage design variables, and build across breakpoints. For content, you can create and update CMS collections and items at scale without opening the editor.

Beyond building, Webflow MCP supports governance and collaboration workflows. Create and manage page branches to iterate safely without affecting your live site, query site activity logs to see what changed and when, and enforce your team's existing roles and permissions so agents only access what they're authorized to touch. You can also analyze site performance conversationally, asking questions about traffic, top pages, and engagement trends, and share that data across your other tools.

The Webflow API is a set of programmatic interfaces that developers use to interact with Webflow data by writing code. Webflow MCP is built on top of those same APIs, but exposes them as tools that an AI agent can use on your behalf — so you don't need to write code to get things done. Describe what you want in plain language, and MCP translates that into the right API calls automatically.

With Webflow MCP, anyone with access to a compatible AI tool can prompt their way through design, content, and site management tasks, while keeping the same underlying data access and security controls the API provides.

Webflow MCP supports parts of a migration or CMS structuring workflow, but it's not built for full site migrations from one platform to another. You can use it to create and manage CMS collections, add and update collection items at scale, and structure content programmatically — which makes it useful for setting up or reorganizing your CMS architecture with the help of an AI agent.

For teams moving content into Webflow or restructuring an existing site, MCP accelerates the process by letting you create collections, populate fields, and manage content through natural language prompts rather than manual data entry. Full one-shot platform migrations are outside what MCP is built for, so it works best as a tool for structured, incremental content work.

Webflow MCP doesn't yet cover every capability available, but its capabilities are actively expanding. Check the developer documentation for the latest on what's supported.

All features

Build and Design
Manage
Optimize
Extend