Fingerprint Flow manipulates your music files (moving, renaming, and rewriting metadata tags). We take data safety extremely seriously.

• Backups before changes: Original files are backed up before any tags are modified (not after). If something goes wrong, the backup contains your original, unmodified file.
• Rollback support: Every file move is recorded in a database. You can undo individual files or entire batch operations.
• Dry-run mode: Preview exactly what would happen before committing any changes.
• Duplicate detection: Files are never silently overwritten. If a destination already exists, the operation is skipped.
• Safe cross-device moves: File integrity is verified (size check) after copying across drives before the source is deleted.
• No source directory deletion: The app never deletes directories outside your configured library path.
• Dangerous path protection: System directories (e.g., C:\Windows, /usr) are blocked as library paths.

1. Always test with a small folder first before processing your entire library.
2. Keep backups enabled (keep_originals: true in config -- this is the default).
3. Use dry-run mode for large batches to preview changes.
4. Set your backup path to a separate drive if possible, so backups survive drive failure.

If you discover a bug that could cause data loss (files deleted, tags overwritten without backup, rollback failure, etc.), please report it as a critical security issue.

1. Do NOT open a public GitHub issue for data-loss bugs -- they may affect users before a fix is available.
2. Email the maintainers at the address listed in the repository's profile, or use GitHub's private vulnerability reporting feature.
3. Include:
   • Steps to reproduce
   • What happened vs. what was expected
   • Your OS, Python version, and Fingerprint Flow version
   • Whether any files were lost or corrupted

• Acknowledgment: Within 48 hours
• Triage: Within 1 week
• Fix: Data-loss bugs are treated as P0 and patched as soon as possible