GitHub connector only shows README when adding private repo to Knowledge Base

polimorfico · March 15, 2026, 11:40pm

I’m trying to set up a Knowledge Base in Dify using a private GitHub repository as the data source, but I’m running into an issue: when I go through the GitHub connector setup, only the README file is available to select. All other files in the repo are not shown.

My setup:

Dify version: 1.13.0
Deployment: self-hosted via Docker
Repo type: private GitHub repository
Token type: Personal Access Token (classic) with repo scope

Steps I followed:

Generated a GitHub PAT with full repo scope
Connected it under Settings → Data Sources → GitHub
Created a new Knowledge Base and selected GitHub as the source
Selected my repo
Only the README.md appears in the file picker — no other files are listed

What I expected:
All files in the repo (.rb, .js, .md, etc.) should be listed and selectable for indexing.

Has anyone run into this before? Is there a file size limit, a folder depth limit, or a specific token permission I’m missing? Any help would be greatly appreciated!

kurokobo · March 16, 2026, 12:21am

As far as I know, that’s a known limitation of the current implementation.

The plugin documentation suggests that code can be retrieved, but that’s a bit of an exaggeration—for now, it seems like only README.md, issues, and pull requests are listed.

github.com/langgenius/dify-official-plugins

datasources/github/datasources/github.py

d6ae6887d


      
          # Add README file (if exists)
          try:
              readme_info = self._make_request(f"{self.base_url}/repos/{repo['full_name']}/readme")
              pages.append({
                  "page_id": f"file:{repo['full_name']}:README.md",
                  "page_name": f"{repo['name']} - README",
                  "last_edited_time": repo.get("updated_at", ""),
                  "type": "file",
                  "url": readme_info.get('html_url', ''),
                  "metadata": {
                      "repository": repo['full_name'],
                      "file_path": "README.md",
                      "size": readme_info.get('size', 0)
                  }
              })
          except ValueError:
              pass  # README doesn't exist

polimorfico · March 16, 2026, 5:54pm

Thanks for your quick reply. It’s quite disappointing that the plugin does not track all repo files Tracking the README file only is unuseful to create a proper knowledge base.