Batch & Selective Export
The Backup page is the heart of XWX Backup: pull conversations from every supported platform into your local vault in bulk, selectively, or on a schedule. Exports are incremental — nothing gets duplicated.
How batch export works
Choose platforms and a count, then save the selected conversations locally in one click.
- Coverage includes all logged-in accounts and conversation types of each platform (multi-account, Projects, Apps)
- The count limit applies to the merged recent conversations across selected platforms
- Already-saved content is skipped automatically — re-running never creates duplicates
- Output formats are configured centrally under Export options
Each item goes: fetch conversation list → fetch full conversation by ID → save to vault → optionally generate PDF/Word output.
Export options & output formats
The vault always saves Markdown — that’s what powers search. Optionally, additional output files are generated per conversation into your chosen export directory:
| Format | Purpose |
|---|---|
| Markdown (.md) | Always saved to vault — searchable, Obsidian-ready |
| Professional layout, generated per conversation | |
| Word (.docx) | Editable document for reports |
If no extra formats are checked, only the vault archive is written — the fastest option for pure backup.
Selective export
For finer control, use the selective export flow:
- Browse the per-platform conversation list (with titles, dates and message counts)
- Filter by account, conversation type, date range, tag or starred status
- Tick individual conversations — or select all filtered
- Click Export selected
This is ideal for topping up specific projects or grabbing conversations you starred earlier.
Pacing & account safety
Bulk reading runs against platform web APIs that weren’t designed for automation, so XWX Backup defaults to conservative behavior:
- ~10 seconds per item by default, with randomized request intervals
- A 2–3 minute cooldown break after every 100 items
- Automatic back-off and retry when a platform returns rate-limit (429) responses
- Per-platform risk levels; platforms owned by the same company are never operated in parallel
You can adjust pacing in Settings (Safer / Balanced / Fast) — slower is safer. Pause and resume mid-run at any time; completed items are remembered, and resuming skips them.
Reading the results & retrying failures
When a run finishes you get a precise summary: added / skipped / failed, plus per-item errors where relevant (e.g. not logged in to a platform).
- Failed items don’t block the rest of the run
- Fix the cause (log in, reconnect the extension), then re-run — previously succeeded items are skipped
- Use Force full re-export if you suspect content changed on the platform and want everything re-fetched and compared