Cloud index and duplicate sources
Mind stores extracted text in the cloud — not a copy of your files. Learn how Skip, Replace, and Allow work when you upload again.
Before you start
- Signed in with a Mind notebook open.
- At least one source already indexed (or upload the same file twice to see dedup).
Steps
- 1
When you upload, FlyType hashes file content (SHA-256 fingerprint) and compares it to sources already in this notebook.
- 2
Skip (default): matching content is not re-uploaded — you keep the existing source id and see a skipped summary in the toast.
- 3
Replace: re-ingest into the same source id with fresh text (use when the file changed on disk).
- 4
Allow: create a new source even when content matches; backend may link duplicate_of for audit.
- 5
Folder and multi-file pickers show an indexed badge when the filename matches an existing source title — deselect or pick a policy before upload.
- 6
Chat and Studio always read indexed text from Postgres — your local folder or cloud drive is not mirrored.
- 7
Origin paths (local path, Drive file id) are metadata for refresh UI only.
Common mistakes
- Re-syncing a project folder? Use Skip to avoid duplicate sources.
- Updated a PDF on disk? Use Replace on that file only.
- Mobile batches show per-file progress and a summary toast with skipped duplicate pairs.
- Large batches and archives — use Hub Mind when mobile caps apply.