GitHub integration use cases
What this actually covers
Before reaching for GitHub import, it's worth being precise about what it does: a single fetch of README.md, the files directly inside a top-level docs/ folder, and the wiki home page, from one public repo and branch you specify. There's no OAuth connection, no scheduled sync, no Issues/PR/code indexing, and no recursive directory walk. Re-running the import with the same repo URL refreshes the content.
That makes it a good fit for a narrower set of cases than a "connect your whole GitHub org" integration would suggest:
1. Turn a project README into chat answers
The most common fit. If your product's setup instructions, CLI usage, or configuration options live in the repo's README, importing it lets the bot answer questions like "how do I install this?" or "what's the config file format?" directly from that content, with a citation back to the GitHub file.
2. Surface top-level documentation
If your docs live as flat Markdown files directly under a docs/ folder at the repo root (not nested in subfolders), those get imported alongside the README. This works well for smaller projects with a handful of top-level guides — architecture overview, getting started, FAQ — rather than a deep, nested docs tree.
3. Wiki-home overview
For repos that use the GitHub Wiki instead of (or alongside) docs/, importing the wiki home page brings in whatever overview content lives there. Only the home page is fetched — other wiki pages aren't followed automatically.
What it does NOT cover
To set expectations correctly against what teams often assume a "GitHub integration" does:
- No Issues or Discussions indexing. The bot can't search closed issues for prior bug reports or answer "has this been reported before?" from GitHub Issues.
- No pull request indexing. PR titles, descriptions, and diffs aren't fetched — the bot can't answer "what did Alice change in PR #1234?"
- No source code indexing. Code files are never fetched, so questions about function internals or "where is X defined?" aren't answerable from this source.
- No nested docs. Files under
docs/adr/,docs/runbooks/, or any other subfolder aren't picked up — only files directly at the top ofdocs/. - No continuous sync. New commits don't trigger re-indexing; you have to re-run the import.
For runbook search, closed-issue lookups, or PR-aware retrieval, the practical option today is exporting that content (e.g. a runbook doc) and bringing it in via file upload or a URL crawl, rather than the GitHub import.
Setup
About a minute:
- Open Knowledge Hub > Add Source > GitHub. See GitHub integration setup.
- Paste the repo URL and branch.
- Check README and/or Wiki home page — the top-level
docs/folder is included automatically when present. - Click Import.
Plan considerations
- Free, Starter, Growth. No GitHub import.
- Business. GitHub import available. Business+
- Enterprise. GitHub import available.
Limits
- Files per import. Up to 50 files from the top-level
docs/directory, plus README and the wiki home page. - Content cap. Imported text counts against your workspace's indexed-content MB limit like any other source.
Common pitfalls
Expected Issues or PR content to show up. It won't — those aren't fetched by this source.
Nested docs files are missing. Only files directly under the top-level docs/ folder are imported; anything in a subfolder (docs/guides/, docs/adr/, etc.) is skipped.
Content looks stale after a new commit. There's no auto-sync — re-run the import to refresh it.
FAQ
Does this work with GitHub Enterprise Server (on-prem)?
No. Import only reaches github.com's public endpoints; there's no configurable base URL or token for a self-hosted instance.
Will the bot leak proprietary code?
No — source files are never fetched in the first place, so there's nothing to leak.
Can I index across organizations?
You can import any public repo you have a URL for, one at a time — there's no org-wide discovery step since there's no OAuth connection.