Source URLs for RAG Citation
When documents are synced to your vector store, Vector Data Loader automatically captures and stores the source URL for each document. This enables your RAG (Retrieval-Augmented Generation) applications to provide citations linking back to the original source.
What is a Source URL?
A source URL is the direct link to the original document in its source system. When users query your vector store and receive results, they can use the source URL to navigate directly to the original content for verification or additional context.
Source URL Formats by Provider
| Source | URL Format | Example |
|---|---|---|
| Confluence | Full page URL | https://yourcompany.atlassian.net/wiki/spaces/TEAM/pages/123456/Page+Title |
| Google Drive | Drive view link | https://drive.google.com/file/d/abc123def456/view |
| Notion | Notion page URL | https://notion.so/workspace/Page-Title-abc123 |
| Website | Original crawled URL | https://docs.example.com/getting-started |
| S3 | Presigned URL | Time-limited secure URL (1 hour expiry) |
| Supabase Storage | Signed URL | Time-limited secure URL (1 hour expiry) |
| File Upload | Signed URL | Time-limited secure URL (1 hour expiry) |
Viewing Storage-Based Documents
Documents from storage-based sources (File Upload, Supabase Storage, S3) use signed URLs for secure access:
- How it works: When you click "Open Document" in the document list or stats page, a time-limited signed URL is generated
- Expiry: Signed URLs are valid for 1 hour after generation
- Privacy: Documents remain private in storage—the signed URL provides temporary, secure access
- Refresh: If a signed URL expires, simply click the link again to generate a new one
Important Notes
- Re-synced Documents: When you re-sync a document, the source URL is preserved from the original sync
- Signed URL Security: Signed URLs provide temporary access without exposing your storage credentials
- Private Sources: Some source URLs may require authentication to access (e.g., private Confluence pages)
- Expired URLs: If you see an error when clicking a storage document link, wait a moment and try again—a fresh signed URL will be generated