Use this workflow before deleting duplicate candidates or normalizing filenames. The aim is a reviewable inventory and proposed name map, not an automatic cleanup.
1. Reproduce the baseline
- Open File Audit Lab and select “Load reproducible demo”.
- Check 1,248 records across 42 folders. These records are synthetic metadata, not files uploaded by users.
- Review the 20 same-size candidate groups and 12 groups with matching demo hashes. Same size alone is not proof of identical contents.
- Download Inventory JSON and keep it as your before snapshot. CSV is useful for reading in a spreadsheet; JSON is the input for Snapshot Compare.
2. Interpret a collision
Consider two synthetic names in the same folder: Report A.txt and report-a.TXT. Lowercasing and converting spaces to hyphens maps both to report-a.txt. The preview must show a collision. Adding a sequence produces distinct proposed names, but a person still decides whether those names are meaningful.
3. Use your own folder
Select only a non-sensitive test folder first. Inventory selection reads metadata; the SHA-256 button reads candidate file contents locally. Keep the 256 MB guard unless you understand the browser memory cost. The application does not delete, move, or rename files. A downloaded rename map is a plan, not an executed change.
What this does not establish
Demo hashes are fixtures. They do not prove the contents of your files. An inventory is not a backup and does not establish that every file is readable or recoverable. A candidate duplicate may still be needed in both locations.
Next step
After making your own separately backed-up changes, export a new inventory and follow the snapshot comparison workflow. Read the hashing guide and collision guide before acting.