Connect GitHub to train your chatbot on documentation that lives in your
repositories: READMEs, docs folders, and other text files.
Connect and sync
Open the connector
In your chatbot’s sidebar, open Files & Data Sources → Add Files
and choose GitHub.
Authorize SiteGPT
Select Connect Account and authorize with GitHub. Private
repositories are available when the connected account has read access to
them.
Point it at the docs
Enter the Repository Owner, Repository Name, and Branch Name,
then add File patterns (globs like **/*.md or docs/**) for the
files the chatbot should learn from, and confirm.
What syncs from GitHub
- Text formats only:
.md, .markdown, .mdx, .txt, .rst, .html,
and .htm.
- Binary files, images, and source code files outside those formats are
skipped.
If your product docs are a static-site repo (Docusaurus, MkDocs, Astro, and
similar), importing the docs folder from GitHub is often cleaner than
crawling the published site: no navigation chrome, no cookie banners.
Keeping it in sync
GitHub is the one connected app with an Auto Sync Frequency option:
choose it when adding the repository and SiteGPT re-imports the matching
files on that schedule (frequencies depend on your plan). You can also
resync manually anytime from Files & Data Sources.
Disconnecting
Revoke the SiteGPT app from your GitHub account
(Settings → Applications). Trained content stays in your chatbot
until you delete it from the list. 
