Connecting a client website touches three things: your role in Accounts, the public website (for AI Memory) and the CMS (for publishing). The crawl needs no login at all. Publishing needs credentials created by someone with admin rights in the CMS. Use this article to collect everything before you start.
In Accounts
To connect a website, or to change a connection later, you must be one of these:
an Owner or Admin of the company, or
an Owner or Admin of the agency the company belongs to.
Anyone else who opens the connection form sees "No access" with the message "Connecting a site is limited to company owners, admins, and agency admins." Starting a crawl needs the same roles.
For AI Memory: public pages only
The crawl reads the public website. It needs no admin login, plugin, DNS change or FTP access.
A sitemap. Ideally listed in robots.txt, or at a standard address such as /sitemap.xml or /sitemap_index.xml. See How the sitemap is retrieved.
No challenge page for automated visitors. Sitemap files are requested with the user agent
UpToDateConnect-Bot/1.0. Page content is fetched through a crawler service, with a direct fetch that uses a regular browser user agent as a fallback. An allow rule for one user agent is therefore not enough if a firewall or bot-protection service challenges unknown visitors. Make sure public pages and the sitemap load without a challenge or consent wall.Rate limits are respected. When the site answers with HTTP 429, the crawler pauses and retries later. Strict limits slow a crawl down; they don't break it.
For publishing: CMS credentials
The sites & blog API (uptodateconnect.com) publishes through the CMS's own API, so it needs credentials created by an admin of that CMS.
CMS | What you need |
|---|---|
WordPress | The site URL, a username or email, and an Application Password for that user. The REST API must answer at |
WordPress, multilingual | Rights to install a plugin. Sites using WPML or TranslatePress need a handler plugin so translations can be linked or pushed. |
Drupal | The site URL, a username and password, and these modules enabled: HTTP Basic Authentication, JSON:API, REST UI, RESTful Web Services and Serialization. JSON:API must allow create operations, and the user must be allowed to create and delete the content type you publish to. |
Webflow | The website domain, the Site ID, the Collection ID of the blog collection and a site API token with CMS read and write access. |
Craft CMS | The site URL, a GraphQL access token that may read, save and delete entries in the blog section, and the section and entry type handles. |
Odoo | The instance URL, the database name, a username or email, an API key for that user, access to Website blog posts and at least one blog to publish to. |
Wix | The website name and URL, the Site ID, an API key with the Wix Blog permission and the blog member who will own the posts. See Connecting Your Wix Site. |
Tip: Create a dedicated CMS user for the platform, for example "UP-TO-DATE publisher". You can then revoke or rotate its access without affecting a real person's login, and posts are clearly attributed.
Checklist before your first onboarding
List the first websites you will connect and the CMS each one runs on.
Arrange admin access to each CMS, or a contact at the client who can create the user, password, token or API key.
Check that each CMS API is reachable from the internet: no IP allow-list, no maintenance mode and no extra password protection in front of the site.
For multilingual WordPress sites, confirm you may install a plugin, and ask the UP-TO-DATE support team for the current handler plugin file.
Confirm your role in Accounts (Owner or Admin of the company, or of your agency).
Keep a contact at the UP-TO-DATE support team for anything that needs platform-side setup, such as turning on TranslatePress support for a site.