Automatic registered-word highlighting works best as an editorial queue, not a bulk-correction command. Separate people, spelling variants, and watch expressions so every match retains the reason it needs review.
The four-chapter sample contains ten PERSON555 matches, six VARIANT555 matches, eight ALERT555 matches, and four intentional exceptions. Stage 4 returned 10, 6, and 8 matches respectively, each across four files, plus four exceptions and zero replacements. Semantic decisions and visual highlighting were not verified, so status is partial. The untouched chapters are the return point.
Keep three lists from becoming one command
The people list checks appearances and forms of address. The variant list compares wording with an editorial standard. The alert list identifies repeated or risky phrasing. Mixing them can turn a proper name into a typo or erase a deliberate expression.
For each match, record list, chapter, surrounding sentence, decision, and any exception reason. “No issue,” “review,” and “individual edit candidate” are enough. Highlighting supplies the queue; context supplies the decision.
Audit 10, 6, 8, and 4
Freeze a copy of the four chapters and the three lists. Review PERSON555, VARIANT555, and ALERT555 one list at a time, preserving chapter distribution. Record which list contains each of the four intentional exceptions and why it stays.
Do not edit while stepping through highlights. Finish the queue first, then move approved items to an individual-edit pass. That preserves the original counts for audit.
Stage 4 reproduced the search queue
PERSON555 returned ten matches, VARIANT555 six, and ALERT555 eight. Every list covered four files. The run recorded four intentional exceptions and replacements=0.
This establishes three separate search sets and an exception set. It does not decide which matches deserve edits or prove the color shown for a selected list. Unsaved tabs were not part of the verified set.
Separate public feature, observation, and limits
Rune Studio’s published feature set allows several registered-word lists and highlights terms from the active list automatically. That is Stage 3. The counts above are Stage 4 search/preview results; visual display remains a limit.
The next desktop run should select one list at a time, trace 10, 6, and 8 visible matches, and confirm that the four exceptions remain unchanged. If unsaved tabs are needed, record them as a separate condition.
Completion means explaining zero replacements
At this stage, success means producing a 24-match review queue while keeping all three list purposes and four exceptions distinct, with zero replacements. If the counts or four-chapter scope change, return to the untouched sample.
STUDIO-548 follows two project-specific lists for name continuity and prose review. This article builds a reusable three-category queue and deliberately ends before replacement. STUDIO-551 and 557 cover font and character count. Check the Rune Studio product page before building the reusable list set.
