You choose the entry type, name column and field mapping before creating draft entries.
Migration is interpretation, not copy and paste.
Mimir needs to understand what incoming information represents before it can become useful structured worldbuilding.
See what is recognised, suggested and left to you.
Mimir can read the table structure. You still select the target entry type, name column and field mapping.
Automation proposes. The creator decides.
Spreadsheet conversion itself is deterministic. An optional Mimir AI step can suggest matches between source columns and your existing fields, but it does not create entries or silently choose your canon.
For prose documents, AI-assisted extraction proposes named and described entities. Every proposal remains reviewable and becomes a follow-up draft only after confirmation.
- Recognised: file type, columns, rows and readable prose
- Suggested: optional column matches or prose entry proposals
- User-selected: entry type, name, mappings and which proposals survive
- Not silently converted: unsupported or unmapped values
Decide where the information belongs.
Choose how a source column becomes a Mimir field, then inspect the draft result before creating entries.
Source: characters.csv
Target entry type: Character
Elara Voss
- Age
- 34
- Tags
- commander · exile
- Summary
This example reflects the verified spreadsheet conversion model. Relationship lists and image-asset fields are not claimed as safe spreadsheet-text mappings.
The uncertain parts should stay visible.
Review concrete conversion limits and choices before they become draft entries. Mimir should not hide uncertainty behind a reassuring progress bar.
Inspect what happened and why.
A spreadsheet row has no usable name.
Why it matters: Mimir needs a name to create an identifiable entry.
Verified behaviour: The row is skipped rather than becoming an unnamed entry.
An imported name already exists.
Why it matters: importing should not overwrite existing material.
Verified behaviour: a unique identifier is generated for the new draft.
A text column targets an image or relationship-list field.
Why it matters: plain spreadsheet text cannot safely recreate those structures.
Verified behaviour: those field types are skipped by the current converter rather than guessed.
AI-assisted prose extraction proposed an entry.
Why it matters: extraction is lossy and a suggestion is not canon.
Your control: rename, re-type or remove the proposal before creating draft entries.
This inspector explains verified conditions and safeguards. It is not a claim that the product currently presents this exact warning interface.
Nothing is created until you choose to create it.
The current flows separate reading and review from the explicit action that creates draft entries.
Follow the information from source to draft.
- 01SourceKeep the original attachment
- 02ReadParse rows or extract text
- 03MapChoose fields or inspect proposals
- 04ReviewEdit, omit or confirm
- 05Create draftsExplicit final action
The original source is present. No Mimir entries have been created.
This is not a promise of a universal dry-run transaction, one-click rollback or perfect recreation of another tool. After import, Mimir's normal local files and history/recovery systems apply to the entries you created.
Move into Mimir without moving into another lock-in.
Once imported entries become part of a Mimir world, they use the same local-first foundation as work created there from the beginning.
The format changes. Your control does not.
Your source remains available as the reference material used for the reviewed conversion.
Ordinary import is not a premium toll.
General import and competitor migration belong to Mimir Core. Local worlds and entries remain readable local files, and downgrading does not erase or lock away that local work.
Imported drafts become ordinary connected entries you can refine.
Use Mimir's normal Git-backed history and recovery after the entries exist.
Standard and safe-to-publish export preserve the principle that your material can leave with you.
Review the mapping. Keep the result.
Explore the wider local-first platform or join the beta and help shape the complete migration workflow.