Skip to main content
Neo Agent reads your Docusaurus site the way a browser does. Once connected, Neo syncs the site’s pages, up to 5,000 a run, into its documentation search once a day, so the Suggest Resolution action and the Find Documentation tool find the runbook that fits a ticket. Docusaurus has no API and no account to connect: Neo needs the site’s address and, for a private site, the sign-in your web server or proxy asks for.

​
Before you connect

  • The site must be reachable from the internet over HTTPS. Neo has no path into your private network. If the site is internal only, publish it through a reverse proxy protected by Basic authentication or a header, as below. You can also limit the proxy to Neo’s IP addresses, but never rely on that alone: every Neo customer’s requests leave from the same addresses, so any of them could connect your site to their own Neo account.
  • Docusaurus itself has no sign-in. If your site is private, it is your web server or proxy that asks for one. Neo can send a Basic authentication username and password, or one header with a secret value (for example X-Api-Key). A sign-in page that needs a person to click (single sign-on) does not work for Neo; give Neo its own way in through the proxy.

​
Connect Docusaurus

1

Add the integration in Neo

In the Neo Dashboard, open Integrations → Documentation and click the Docusaurus tile.
  1. Site URL: the address of the site’s home page, as it appears in the browser. Include the base path if the site has one, for example https://example.com/kb.
  2. Sign-in: choose No sign-in only for a site anyone may read, Basic authentication to enter a username and password, or A header your proxy checks to enter the header’s name and value.
Neo opens the home page before it saves, and tells you if the site refused the sign-in, redirected elsewhere, or is not a Docusaurus site.
2

Prove the site is yours

The first save asks you to add a file named neo-agent-verification.txt to your site’s static folder, holding one line that starts with neo-agent-site-verification=. The line is unique to your Neo account, and the save message shows it. Docusaurus copies the static folder to the site root, so after you build and deploy the site, the file is at <site URL>/neo-agent-verification.txt. Then save again.Neo checks the file at every save and before every sync. It stops anyone else’s Neo account from reading your site, even one that reaches it through the same proxy. If you remove the file, syncs stop and Neo tells you in your inbox.
3

The first sync starts on its own

Saving the site switches the Docusaurus sync on. The first run starts on Neo’s next scheduled sync check, and syncs then run daily. The Docusaurus card on the Integrations page shows the last run’s status, and Trigger sync starts a run straight away, unless one is already running. Once it finishes, your pages are searchable through Find Documentation and reach the Suggest Resolution action next to every other documentation source you connected.

​
What gets synced

Neo reads the site’s sitemap.xml, which Docusaurus’ classic preset writes. A site built with noIndex: true has no sitemap, so Neo starts at the home page and follows the site’s links instead; every page carries the sidebar, so Neo still reaches every doc. A page with less than 100 characters of text, title included, is left out. Each synced page carries its link on your site. Your docs site is your own, not one client’s, so every page is available to every company’s tickets. Pages from a site behind Basic authentication or a header are marked confidential: your end users’ chat agents never see them, and your technicians’ agents read them marked as confidential. Pages from a public site are available to both. Neo re-indexes a page when it changes. It reads the change date from the page’s Last updated line (showLastUpdateTime), the sitemap’s lastmod, or your web server; a page with none of these is indexed again on each run. A run reads at most 5,000 pages, downloads at most 512 MB, spends at most 20 minutes reading page content and indexes at most 100 MB of text, and a page beyond any of these is not synced. While a site stays above that, Neo keeps the pages it synced earlier that a run did not reach; disconnecting Docusaurus removes every synced page.

​
Disconnect Docusaurus

Open the Docusaurus tile and choose Disconnect. Neo deletes the site address and sign-in, stops the sync and removes the synced pages from its search, since nothing would keep them current. Your site is not touched. To get the pages back, connect Docusaurus again and the next sync indexes them.

​
Troubleshooting

Check the username and password, or the header name and value, on the Docusaurus tile. If a firewall in front of the site also filters by address, add Neo’s IP addresses there.
Check that <site URL>/neo-agent-verification.txt opens in a browser and holds the exact line the save message shows, then save again. The file must be in the static folder of the Docusaurus project, and the site must be built and deployed after you add it.
Enter the URL of the site’s home page, with its base path if it has one. A page in front of the site, such as a sign-in page, also gives this answer: give Neo a way past it with Basic authentication or a header.
The site may render its pages in the browser only (Docusaurus’ hash router), which leaves no text for Neo to read. Build the site with the default router.