Choose a text editor for novel-to-EPUB work by testing content authority, not by counting features or applications. Start from NOVEL-SOURCE-01, which has three chapters, five recorded character-name locations, two illustration references, three navigation entries, and one known C02 typo. Generate novel-v1.epub, correct only the known typo in the authority, generate novel-v2.epub, and reject unrelated changes.
Use the application-handoff selection method when the decision hinges on minimizing application boundaries for a one-application trial. This article allows more than one tool when the source-and-asset relationships remain traceable. Text generation, collaboration, KDP review, and price ranking are excluded.
Decision: preserve source and asset relationships across two packages
Run these six stages in order:
- identify the authoritative three-chapter source;
- reconcile five prepared character-name locations;
- map two source illustration references to package assets;
- verify three navigation labels and destinations;
- preserve
novel-v1.epub, then correct one known C02 typo in the authority; - compare
novel-v2.epubwith the ledger and the first package.
Any unverified stage keeps the candidate pending. Strong search cannot offset a missing illustration relationship.
Freeze the authority and asset ledger
Name C01 through C03 collectively as NOVEL-SOURCE-01 and allow only one editable authority. Record five character-name anchors, source references for illustration-01.jpg and illustration-02.jpg, three navigation labels and destinations, and the known typo ID.
Give every candidate the same copy. Reducing the chapter or image count for one candidate changes the novel condition rather than measuring the workflow.
Reconcile names, illustrations, and navigation
For the character name, confirm locations and surrounding passages, not only a total count. For each illustration, connect the manuscript reference, source asset, and packaged asset in one ledger row.
Do not assume a renamed image updates its references automatically. Record the explicit repair operation and excluded targets such as external URLs or page anchors.
For navigation, verify both each displayed label and its chapter destination. Correct order with a wrong destination is not a pass.
Compare the first package with the ledger
Identify the first result as novel-v1.epub. Reconcile C01 through C03, five name locations, two illustrations, and three navigation entries with the ledger. Keep verified and unverified rows separate.
“Export succeeded” does not fill these rows. Connect source ID, package ID, and package check time.
For every evidence row, retain the source reference location, packaged asset path, result, and return destination. A visible illustration or navigation item alone cannot identify which source and which package were checked.
Correct one known typo only in the authority
Return to the recorded C02 typo and correct that one item in NOVEL-SOURCE-01. Do not patch generated XHTML or the first EPUB. Preserve the first package.
Confirm the source diff is limited to the intended item, then generate novel-v2.epub under the same chapter, navigation, and illustration settings.
Separate allowed and forbidden differences
Allowed differences are the known C02 correction and necessary artifact identity or time metadata. Forbidden differences include other prose, five name locations, two illustration relationships, navigation labels and destinations, and chapter order.
Classify differences by ledger row rather than by count. An illustration-path or navigation-order change returns to the source relationship or settings that produced it.
State the partial Rune Studio evidence
Rune Studio documentation describes long-form editing, manuscript organization, search, image and link notation, preview, chapter and navigation selection, and EPUB output in one workspace. That makes it a candidate for maintaining this ledger.
Hands-on checks separately covered search and replacement, explicit reference rewriting, image-free EPUB generation and package check, and one-item regeneration on common specimens. Rename alone did not update references; an explicit rewrite was required. Cover-and-body-image package check also remains affected by the current package checker discrepancy.
The three chapters, five name locations, two illustrations, three navigation entries, and two packages have not completed one article-specific route. Rune Studio therefore remains pending on this exact ledger.
Conclusion: choose from novel invariants, not handoff count
A novel editor candidate advances when one authority, five name locations, two illustrations, and three navigation destinations remain traceable across two packages, with only the planned C02 correction changing.
For your own test, generate the first package and record all five character-name locations, the stored path and reference for each illustration, and the three navigation labels and destinations. Then correct only the known C02 typo and generate a second package under a new identity. Require the old spelling to appear zero times and the new spelling once, while the other four name locations, both illustration relationships, and all three navigation destinations still match the first package. If one relationship cannot be explained, reject the second package and keep the first as the accepted baseline. Return to the authoritative source and asset ledger, repair the mismatched row, and repeat every invariant rather than checking only the repaired item. Record both package identities and the comparison time so a later editor can reproduce which package passed. This recovery route distinguishes a traceable novel build from an output that merely opens without showing whether its names, images, and navigation came from the intended source.
Begin with NOVEL-SOURCE-01 and the asset ledger. If Rune Studio is a candidate, review the current Mac scope on the Rune Studio product page and do not promote the partial hands-on testing checks into a completed novel package route.