Webflow Integration

The Webflow integration connects a Webflow site to ReachLLM's Content & CMS adapter system.

What it can do

After connection, ReachLLM can:

• List and read Webflow static pages.
• Return editable static page nodes from the Webflow Data API so the agent can preserve IDs and update specific text/component content.
• List, read, create, and update Webflow CMS items.
• Publish generated articles into a Webflow CMS collection.
• Upload images into Webflow assets when the token has asset write access.
• Carry generated article images in the rich text body, with optional hero image upload and mapping into a matching collection image field.
• Update metadata for supported Webflow pages and CMS items.
• Append schema to editable CMS item content.
• Create or update /llms.txt as a Webflow well-known file when the token and workspace support site configuration writes.
• Create an llms.txt CMS item fallback when root well-known files are not available and a CMS collection is available.
• Update /robots.txt through Webflow site configuration on supported Enterprise workspaces.

Webflow's public API does not expose a safe full-body replacement for every static page. For static pages, ReachLLM can read content, expose editable node IDs, update metadata, and send targeted node updates. Full redesigns should target CMS items, targeted editable page nodes, or be completed in Webflow Designer.

Webflow targeted page updates use the Webflow Data API static page DOM endpoints. If the target is a localized page, Webflow requires a secondary locale_id. Keep the original node IDs, component property IDs, and any data-w-id attributes returned by cms_get_content when sending a node update.

Root file updates use Webflow Enterprise site configuration APIs when available. Webflow accepts structured robots rules, not arbitrary raw robots.txt text, so ReachLLM applies User-agent, Allow, Disallow, and Sitemap directives and reports unsupported directives in the result. If /llms.txt cannot be written as a well-known file, ReachLLM falls back to publishing an llms-txt CMS item.

How to connect

1. Open Account > Integrations.
2. Go to Content & CMS.
3. Click Connect on Webflow.
4. Enter a Webflow Data API token with Site and CMS read/write access.
5. Optionally enter the Webflow Site ID if the token can access multiple sites.
6. Optionally enter the Blog Collection ID if you want article drafts to use a specific CMS collection.
7. Click Connect Webflow.

If you connect through the Composio/API-key form, the Site ID, Site URL, and Blog Collection ID fields are ReachLLM-only targeting hints. ReachLLM stores them with the CMS mirror so publishing and retrieval use the intended Webflow site and collection.

ReachLLM stores the API token server-side. The brand profile only keeps connection status, site metadata, and capabilities.

Agent workflow

When you ask the agent to update Webflow content, it should:

1. Run cms_get_status.
2. Run cms_list_content to find the target page or CMS item.
3. Run cms_get_content to read the current content.
4. Use cms_update_content for CMS items or supported targeted page-node updates.

For CMS items, the agent can pass a Webflow collection ID, content field slug/API name, custom field data, draft/publish status, slug, excerpt, and hero image URL when the collection schema supports those fields.

When the user asks to add or replace a picture, the agent can call cms_upload_media first. That uploads the image to Webflow assets and returns a URL that can be reused in cms_update_content or mapped into a collection image field. If asset upload is unavailable for the token, article publishing falls back to the original generated image URL.

For static pages, the agent should only send targeted nodes from the editableNodes returned by cms_get_content; it should not attempt unsafe full-page body replacement. For localized static pages, pass locale_id with the node update. If the node belongs to a branch, pass branch_id as well.

For generated articles, the agent can publish directly into the configured CMS collection as a draft or published item.

Website Generator source extraction

When Webflow is connected, Website Generator source discovery asks Webflow for pages and CMS items before using sitemap results. If you ask ReachLLM to extract or redesign a current page, the source response can come directly from Webflow and includes the platform, content ID, type, URL, title, body content, metadata, and editable node details when available.

For Webflow static pages, source extraction is read-first and node-aware. The agent should use the returned editable node IDs for targeted edits. For CMS items, the agent can use the CMS item ID to update rich text, metadata, schema, and article content through the adapter.