> ## Documentation Index
> Fetch the complete documentation index at: https://sitegpt.ai/docs/llms.txt
> Use this file to discover all available pages before exploring further.

# GitHub

> Train your chatbot on Markdown docs, READMEs, and files from GitHub repositories

# GitHub

Connect a GitHub repository to train your SiteGPT chatbot on the content in your codebase, such as Markdown documentation, READMEs, and other text-based files. This is ideal for developer docs, internal guides, and any knowledge that already lives alongside your code.

## Prerequisites

* A GitHub account (personal or organization)
* Access to the repository you want to import
* Owner or Editor permissions on the SiteGPT chatbot

## Connecting GitHub

<Steps>
  <Step title="Navigate to Training">
    Go to your chatbot dashboard and open the **Training** tab.
  </Step>

  <Step title="Add Data Source">
    Click **Add Files** and select **GitHub**.
  </Step>

  <Step title="Authenticate">
    Click **Connect GitHub** and sign in with your GitHub credentials. Grant SiteGPT permission to read the repositories you want to import from.
  </Step>

  <Step title="Select Content">
    Choose the repository and branch, then select the files or folders you want to import. You can pick an entire folder (such as a `docs/` directory) or individual files.
  </Step>

  <Step title="Start Training">
    Click **Import Selected** to begin training. SiteGPT processes each file and adds the content to your chatbot's knowledge base.
  </Step>
</Steps>

## Supported File Types

GitHub repositories are best suited for text-based content. SiteGPT can process:

| File Type | Extensions           |
| --------- | -------------------- |
| Markdown  | .md, .markdown, .mdx |
| Documents | .txt, .rst           |
| Web       | .html, .htm          |

<Note>
  Binary files (images, archives, compiled assets) are skipped. For best results, point SiteGPT at the documentation files in your repository rather than the entire source tree.
</Note>

## Best Practices

### Point at your docs

Most repositories keep documentation in a dedicated folder. Select that folder (for example `docs/`) so your chatbot trains on the content meant for readers, not build scripts or configuration files.

### Use the right branch

Import from the branch that holds your published documentation (often `main`). Re-import after merging significant doc changes to keep answers current.

### Exclude noise

Skip generated output, changelogs you do not want surfaced, and test fixtures. The cleaner the selection, the more focused your chatbot's answers will be.

## Troubleshooting

<AccordionGroup>
  <Accordion title="Repository or files not appearing">
    * Verify you granted SiteGPT access to the correct repository during authentication
    * Private repositories require that the connected account has read access
    * Check that the files are in a supported text-based format
  </Accordion>

  <Accordion title="Import fails or times out">
    * Try importing a smaller folder or fewer files at once
    * Very large repositories may take longer to scan
    * Confirm your GitHub connection is still authorized
  </Accordion>

  <Accordion title="Content looks out of date">
    * Re-import after pushing changes; imports capture a snapshot at import time
    * Make sure you imported from the branch that contains your latest docs
  </Accordion>
</AccordionGroup>