v1.6.0

Changelog - Version 1.6.0

History-Aware Tracking

Version 1.6.0 introduces History-Aware Tracking for automatic undo/redo management of chapter summaries, along with important bug fixes.

---

New Features

History-Aware Tracking

Automatically manages chapter summaries when you use undo/redo operations. No more orphaned summaries or missing entries after navigating your story history!

How It Works:

• When you undo past a chapter break: The orphaned summary is automatically removed from the lorebook
• When you redo to restore a chapter break: The cached summary is restored from history storage
• Works with undo, redo, retry, and history jumps

Technical Details:

• Uses NovelAI's onHistoryNavigated hook to detect navigation events
• Stores complete chapter state (summaries, fingerprints, condensed ranges) per history node
• Validates cached state against actual document to prevent stale state restoration
• Full state preservation through condensation operations

Configuration:

• Key: history_aware_tracking
• Type: boolean
• Default: true (enabled by default)
• Label: "History-Aware Tracking"

---

Bug Fixes

Rebuild Counting In-Progress Chapter

Symptom:
"Rebuild All Summaries" was counting the currently in-progress chapter when calculating chapters to rebuild. With 3 chapter break tokens, it would try to summarize 4 chapters instead of 3.

Root Cause:
The analysis used chapterBreakCount + 1 which incorrectly included the text after the last break as a "complete" chapter.

Solution:
Changed to only process complete chapters (those with an ending break). The number of complete chapters equals the number of break tokens.

Backup Failing on New Stories

Symptom:
On a newly created story with no content, creating a backup would fail with:

Backup failed: cannot read property 'length' of undefined

Root Cause:
The failedChapters storage retrieval was missing a fallback value, returning undefined on new stories.

Solution:
Added || [] fallback to failedChapters retrieval in createRebuildBackup().

Stale Lorebook Entry IDs After Redo

Symptom:
After using redo to restore chapter summaries, uncondense operations would fail or target wrong entries because the condensedRanges metadata contained old entry IDs.

Root Cause:
When restoreFullHistoryState() recreated lorebook entries, they received new IDs from NovelAI. The cached condensedRanges metadata still referenced the old entry IDs.

Solution:
After recreating entries, restoreFullHistoryState() now:

1. Queries back all newly created entries to get their IDs
2. Builds a map from display name to new entry ID
3. Updates condensedRanges with the correct entry IDs before saving to storage

---

Technical Changes

New Types

// v1.6.0: History-aware tracking types
type HistoryChapterState = {
    chapterBreakCount: number;
    summaries: CachedChapterSummary[];
    condensedRanges: CondensedRange[];
    fingerprints: ChapterFingerprint[];
    lastProcessedChapterCount: number;
    timestamp: number;
}

type CachedChapterSummary = {
    chapterNumber: number;
    title: string;
    summaryText: string;
    lorebookEntryId: string;
    isCondensed: boolean;
    startChapter?: number;
    endChapter?: number;
}

New Functions

Function	Purpose
storeHistoryChapterState()	Stores complete chapter state to current history node
getHistoryChapterState(nodeId)	Retrieves cached state for a specific history node
countCurrentChapterBreaks()	Counts chapter breaks in current document
restoreFullHistoryState(state)	Performs complete state restoration from cache
historyStatesAreDifferent(a, b)	Compares two history states for changes
onHistoryNavigatedHook(params)	Main handler for history navigation events

New Configuration Variables

let historyAwareTracking: boolean;  // Default: true

Hook Registrations

// History-aware tracking hook
api.v1.hooks.register('onHistoryNavigated', onHistoryNavigatedHook);

---

Changes Summary

Change	Type	Description
History-Aware Tracking	Feature	Auto-manage summaries on undo/redo
Rebuild Chapter Count	Bug Fix	No longer counts in-progress chapter
Backup on New Stories	Bug Fix	Fixed undefined length error
Stale Entry IDs After Redo	Bug Fix	Updates condensedRanges with correct entry IDs
Stale Cache Detection	Enhancement	Validates cached state matches document

---

Migration from v1.5.6

For Users

No action required! Simply update to v1.6.0:

1. Copy the new script code
2. Paste into NovelAI script editor
3. Save and reload

All your existing data remains compatible:

• Fingerprints preserved
• Condensed ranges intact
• Backups fully compatible
• Settings unchanged

New Configuration Options

After updating, you can configure the new feature in NovelAI's script configuration:

• History-Aware Tracking - Enabled by default
  • Disable if you prefer manual summary management after undo/redo

---

Version History

Version	Date	Type	Description
v1.6.0	2025-11-29	Feature	History-aware tracking
v1.5.6	2025-11-27	Bug Fix	Status panel initialization fix
v1.5.5	2025-11-25	Maintenance	API compatibility (modal async)
v1.5.4	2025-11-24	Feature	Permissions system integration
v1.5.3	2025-11-22	Feature	Manual condensation controls

---

Known Limitations

1. Pre-Tracking History: History nodes that existed before the script was loaded will not have cached state. Undoing to these nodes syncs lorebook to document state.

---

Support

Issues or Questions?

• GitHub Issues: Report problems or ask questions
• Documentation: See USER_MANUAL.md for feature documentation

Reporting Problems

If you encounter issues:

1. Enable DEBUG_MODE in the script
2. Check browser console for [History] prefixed messages
3. Note exact symptoms and reproduction steps
4. Report on GitHub with details and log output

---

Version: 1.6.0
Release Date: November 29, 2025
Type: Feature Release
Previous Version: 1.5.6

v1.5.6

Changelog - Version 1.5.6

Status Panel Initialization Fix

Version 1.5.6 fixes a critical bug where the Status Panel would remain stuck on "Loading status...", "Loading...", "Loading..." indefinitely, even after the script successfully initialized.

---

Bug Fixed

Status Panel Stuck on "Loading..."

Symptom:
After script initialization, the Status Panel displayed:

• "Loading status..."
• "Loading..."
• "Loading..."

Despite logs showing successful initialization:

✓ Permissions check: All granted
✓ Loaded category ID from storage
✓ Found 0 chapter entries in lorebook
✓ Initial status panel update attempt 1 completed

Root Cause:
The updateParts API silently fails during panel initialization. The previous approach registered the panel with placeholder content, then attempted to update it immediately afterward. Even with retry loops, the updates were silently ignored by NovelAI's UI system.

Solution:
Build real content before registering the panel, so the panel is registered with actual data instead of placeholders.

---

Technical Changes

New Helper Function: buildStatusDisplayText()

Extracted status text building logic into a reusable helper function that can be called during initialization (before panel registration) as well as during normal updates.

async function buildStatusDisplayText(): Promise<string> {
    // Builds complete status text with:
    // - Token usage
    // - Background work indicators
    // - Auto-detection notifications
    // - Summary breakdown
    // - Failed chapters list
}

Pre-Registration Content Building

The initialization now builds real content for all three dynamic boxes before registering the panel:

// v1.5.6: Build real content BEFORE panel registration
let initialStatusText = "**Initializing...**\n\n...";
let initialChangedChaptersContent: UIPart[] = [...];
let initialCondensedRangesContent: UIPart[] = [...];

// Try to build real content if we have permissions
if (hasRequiredPermissions && lorebookCategoryId) {
    try {
        initialStatusText = await buildStatusDisplayText();
        initialChangedChaptersContent = await buildChangedChaptersUI();
        initialCondensedRangesContent = await buildCondensedRangesUI();
    } catch (error) {
        // Fall back to placeholder content
    }
}

// Register with pre-built content
await api.v1.ui.register([{
    type: "scriptPanel",
    content: panelContent  // Contains pre-built content
}]);

Permission Check API Workaround

Changed checkPermissions() to use list() instead of has() due to a NovelAI API bug where has() returns false even when permissions are granted.

// v1.5.6: Using list() instead of has() due to NovelAI API bug
const currentPerms = await api.v1.permissions.list();
const hasPerms = REQUIRED_PERMISSIONS.every(perm => currentPerms.includes(perm));

Startup Timing Resilience

Added retry logic during startup to handle timing issues where NovelAI's permission system may not be ready immediately when the script loads:

// v1.5.6: Check permissions with retry to handle timing issues
for (let attempt = 0; attempt < 3; attempt++) {
    await checkPermissions();
    if (hasRequiredPermissions) {
        await initializeLorebookCategory();
        break;
    }
    await api.v1.timers.sleep(300 * (attempt + 1));
}

updateStatusPanel() Resilience

Added retry logic in updateStatusPanel() to re-check permissions and initialize the lorebook category if needed, handling cases where initialization was incomplete:

// v1.5.6: Re-check permissions and initialize if needed
if (!hasRequiredPermissions || !lorebookCategoryId) {
    for (let attempt = 0; attempt < 3; attempt++) {
        // Retry permission check and category initialization
        // ...
    }
}

---

Changes Summary

Change	Description
Bug Fix	Status panel no longer stuck on "Loading..."
New Function	buildStatusDisplayText() helper for reusable status building
Init Change	Content built before panel registration
API Workaround	Using list() instead of has() for permission checks
Retry Logic	Added startup timing resilience with retries
Simplified	Removed ineffective retry loop after panel registration

---

Migration from v1.5.5

For Users

No action required! Simply update to v1.5.6:

1. Copy the new script code
2. Paste into NovelAI script editor
3. Save and reload

All your existing data remains compatible:

• Fingerprints preserved
• Condensed ranges intact
• Backups fully compatible
• Settings unchanged

Verification

After updating, the Status Panel should immediately display:

• Token usage statistics
• Summary breakdown
• Any failed chapters
• Changed chapters list (if any)
• Condensed ranges (if any)

If you see actual content instead of "Loading...", the fix is working.

---

Version History

Version	Date	Type	Description
v1.5.6	2025-11-27	Bug Fix	Status panel initialization fix
v1.5.5	2025-11-25	Maintenance	API compatibility (modal async)
v1.5.4	2025-11-24	Feature	Permissions system integration
v1.5.3	2025-11-22	Feature	Manual condensation controls

---

Technical Details

Why updateParts() Was Failing

The NovelAI updateParts API has specific behavior during initialization:

1. Silent Failure: updateParts does not throw errors when it cannot find elements or when called too early
2. Timing Dependency: Panel elements may not be queryable immediately after registration
3. No Feedback: Success or failure is not communicated back to the caller

The previous retry loop:

// OLD APPROACH (v1.5.5) - This silently failed
for (let attempt = 1; attempt <= 3; attempt++) {
    await api.v1.timers.sleep(100);
    await updateStatusPanel();  // Silent failure
}

The new approach eliminates the need for post-registration updates by ensuring the panel is registered with complete, accurate content from the start.

Files Modified

• Automatic_Chapter_Summaries_v1.5.6.ts - Main script file

Code Locations

• buildStatusDisplayText(): Lines 5968-6063
• Pre-registration content building: Lines 6949-6978
• Permission check workaround: Lines 321-340
• Startup retry logic: Lines 7005-7015
• updateStatusPanel() resilience: Lines 6137-6175

---

Support

Issues or Questions?

• GitHub Issues: Report problems or ask questions
• Documentation: See main README.md for feature documentation

Reporting Problems

If you still encounter initialization issues after updating:

1. Enable DEBUG_MODE in the script
2. Check browser console for error messages
3. Note exact symptoms (what displays vs what should display)
4. Report on GitHub with details and log output

---

Version: 1.5.6
Release Date: November 27, 2025
Type: Bug Fix
Previous Version: 1.5.5

v1.5.5

Changelog - Version 1.5.5

API Compatibility Update

Version 1.5.5 updates the script to be compatible with breaking changes in the NovelAI Script API. This is a maintenance release with no new features - all functionality remains identical to v1.5.4.

---

🔧 API Breaking Changes Addressed

Modal API Updates

The NovelAI Script API introduced breaking changes to the modal system:

Before (v1.5.4 and earlier):

const modal = api.v1.ui.modal.open({...});  // Returns ModalHandle directly
modal.close();  // Synchronous
await modal.closed;  // Wait for close

After (v1.5.5):

const modal = await api.v1.ui.modal.open({...});  // Returns Promise<ModalHandle>
await modal.close();  // Asynchronous
await modal.closed;  // Wait for close

---

📝 Changes Made

All Modal Operations Updated

• ✅ 17 modal.open() calls - All now use await to retrieve the modal handle
• ✅ All modal.close() calls - All now use await as the method is async
• ✅ Callback functions - Made async wherever they use await operations
• ✅ Promise executors - Made async for 3 modal creation patterns
• ✅ .then() patterns - Converted modal.closed.then() to await modal.closed

Updated Patterns

Modal Creation:

// Before
const modal = api.v1.ui.modal.open({
    title: "My Modal",
    content: [...]
});

// After  
const modal = await api.v1.ui.modal.open({
    title: "My Modal",
    content: [...]
});

Modal Closing:

// Before
callback: () => {
    modal.close();
}

// After
callback: async () => {
    await modal.close();
}

Chained Operations:

// Before
modal.close();
modal.closed.then(() => {
    showNextModal();
});

// After
await modal.close();
await modal.closed;
showNextModal();

---

🎯 Affected Features

All modal-based features have been updated:

User Interface Modals

• ✅ Condensed range details viewer
• ✅ Backup details browser
• ✅ Backup restore confirmation
• ✅ Changed chapters notifications
• ✅ Token budget warnings

Configuration Modals

• ✅ Condensation settings modal
• ✅ Manual condensation range selector
• ✅ Generation limit warnings
• ✅ Confirmation dialogs

All Modals Now Properly Handle

• Async opening (awaits modal handle)
• Async closing (awaits close completion)
• Async callbacks (properly marked as async)
• Proper await patterns throughout

---

🔄 Migration from v1.5.4

For Users

No action required! Simply update to v1.5.5:

1. Copy the new script code
2. Paste into NovelAI script editor
3. Save and reload

All your existing data remains compatible:

• ✅ Fingerprints preserved
• ✅ Condensed ranges intact
• ✅ Backups fully compatible
• ✅ Settings unchanged

For Developers

If you've modified the script, update your modal patterns:

1. Add await before all modal.open() calls
2. Add await before all modal.close() calls
3. Make callbacks async if they use await
4. Convert .then() chains to await patterns

---

🧪 Testing Performed

Comprehensive Modal Testing

• ✅ All 17 modal.open() patterns verified
• ✅ All modal.close() operations tested
• ✅ Callback async conversions validated
• ✅ No TypeScript compilation errors
• ✅ All Promise patterns updated correctly

Feature Testing

• ✅ Condensed range details modal
• ✅ Backup browser and restore
• ✅ Changed chapters UI interactions
• ✅ Manual condensation modals
• ✅ Generation limit warnings
• ✅ Token budget notifications

---

📊 Technical Details

Files Modified

• Automatic_Chapter_Summaries_v1.5.5.ts - Main script file

Code Changes Summary

• 17 locations: Added await before api.v1.ui.modal.open()
• ~25 locations: Added await before modal.close() calls
• ~15 callbacks: Made async to support await usage
• 3 Promise executors: Made async for modal operations
• ~5 patterns: Converted .then() to await style

Pattern Examples

Simple Modal:

- const modal = api.v1.ui.modal.open({
+ const modal = await api.v1.ui.modal.open({
      title: "Example",
      content: [...]
  });

Modal in Callback:

  callback: () => {
-     modal.close();
-     modal.closed.then(() => {
-         showNext();
-     });
+     await modal.close();
+     await modal.closed;
+     showNext();
  }

Note: Callback also needs async keyword

Promise Pattern:

- return new Promise((resolve) => {
-     const modal = api.v1.ui.modal.open({
+ return new Promise(async (resolve) => {
+     const modal = await api.v1.ui.modal.open({
          title: "Confirm",
          content: [...]
      });

🔮 Future Compatibility

NovelAI API Tracking

This release addresses the modal API changes announced on November 25, 2025. Future API changes will be tracked and addressed in subsequent releases.

Related API Changes (Not Yet Implemented)

The NovelAI API changelog also mentioned:

• ✅ buildContext parameter renamed to contextLimitReduction - Not used by this script
• ✅ update/close methods removed from type declarations - Already using instance methods
• ✅ createCancellationSignal drain fix - Not used by this script
• ✅ onBeforeContextBuild hook added - Not currently needed
• ✅ onTextAdventureInput hook added - Not currently needed
• ✅ getOrDefault undefined handling fix - Not affected

---

📈 Version History Context

Recent Versions

• v1.5.5 (2025-11-25) - API compatibility update (this release)
• v1.5.4 (2025-11-24) - Permissions system integration
• v1.5.3 (2025-11-22) - Manual condensation controls
• v1.5.2 (2025-11-22) - Fixed automatic processing bypass
• v1.5.1 (2025-11) - Condensed ranges UI

Upgrade Path

v1.5.4 → v1.5.5: Just update the script (no data changes)
v1.5.3 → v1.5.5: Update and benefit from permissions + API fixes
v1.5.x → v1.5.5: All data fully compatible

---

🎯 Impact Assessment

User Impact

Minimal - Users will see no difference in functionality:

• All features work identically
• No UI changes
• No data format changes
• No configuration changes required

Developer Impact

Low - Simple pattern updates required:

• Straightforward async/await conversions
• No complex refactoring needed
• TypeScript catches all issues at compile time
• Clear error messages for missed patterns

---

🔍 Verification

How to Verify the Update

1. Check Version Number

   // In script header
   // Chapter Summarize v1.5.5: API compatibility update

2. Check Changelog

   // CHANGELOG v1.5.5:
   // - [API COMPATIBILITY] Updated for NovelAI Script API breaking changes

3. Test Modal Operations

   • Open "View/Restore Backups" - should work smoothly
   • Open "Condense With Settings" - should work smoothly
   • Try any modal-based feature - all should function normally

---

📞 Support

Issues or Questions?

• GitHub Issues: Report problems or ask questions
• Documentation: See main README.md for feature documentation
• API Changes: Refer to NovelAI's API changelog for details

Reporting Problems

If you encounter issues after updating:

1. Enable DEBUG_MODE in the script
2. Check browser console for error messages
3. Note which modal/feature is affected
4. Report on GitHub with details

---

✅ Summary

Version 1.5.5 is a compatibility update that ensures the Automatic Chapter Summaries script works with the latest NovelAI Script API changes. All modal-based operations have been updated to use the new async patterns.

Key Points:

• ✅ All 17 modal operations updated
• ✅ Full async/await compliance
• ✅ No data format changes
• ✅ No feature changes
• ✅ No user action required
• ✅ Fully tested and validated

Recommendation:
All users on v1.5.4 should update to v1.5.5 to ensure continued compatibility with NovelAI's platform.

---

Version: 1.5.5
Release Date: November 25, 2025
Type: Maintenance / API Compatibility
Previous Version: 1.5.4
Next Version: TBD

v1.5.2

Changelog - Version 1.5.2

Release Date

November 22, 2025

Overview

Version 1.5.2 is a major stability and control release that fixes critical bugs in automatic processing behavior. The script now properly respects user configuration settings and provides three distinct behavior modes based on the autoDetectOnGeneration and autoRegenerate settings.

---

Critical Bug Fixes

1. Automatic Processing Bypass Fixed

Problem: Script was processing all chapters automatically when pasting multi-chapter stories, completely ignoring the autoDetectOnGeneration and autoRegenerate configuration settings.

Solution:

• Removed automatic multi-chapter batch processing loop from onResponse hook
• Added proper configuration checks before any automatic processing
• Script now respects all three behavior modes (see Configuration Modes below)

Impact: Users now have full control over when and how chapter summaries are generated.

---

2. Generation Counter Limit During Condensation

Problem: When condensing chapters, the script would hit the "Non-interactive generation limit reached (5 calls)" error because the generation counter wasn't being reset between different operations.

Solution:

• Reset generationCounter = 0 at the start of autoRegenerateChanges() batch processing
• Reset counter before Level 1 (normal) condensation
• Reset counter before Level 2 (aggressive) condensation
• Reset counter before Level 3 (emergency) condensation

Impact: Condensation now works reliably without hitting generation limits, even during aggressive multi-level condensation scenarios.

---

3. Wrong Chapter Generation on Story Paste

Problem: When pasting a story with 10 chapter breaks, the script would generate "Chapter 10" first using Chapter 1's text content.

Solution:

• Skip single-chapter generation in full auto mode when no fingerprints exist
• Let the automatic batch detection handle all chapters at once
• Only process single chapters when appropriate based on configuration mode

Impact: Pasted multi-chapter stories are now processed correctly with proper chapter numbering and content.

---

4. Chapter Numbering Off-By-One Error

Problem: Script was determining chapter readiness incorrectly, requiring N+1 breaks for chapter N to be ready.

Solution:

• Changed chapterToSummarize calculation from chapterBreakCount - 1 to chapterBreakCount
• Logic now correctly recognizes: N breaks = N complete chapters, chapter N+1 in progress
• Chapter N is ready to summarize when the Nth break exists

Impact: Chapters are now detected and processed at the correct time based on break count.

---

5. Incomplete Chapters Being Detected

Problem: When no fingerprints existed, the script would detect chapters that were still in progress (not yet complete).

Solution:

• Modified detectChangedChapters() to only detect complete chapters when no fingerprints exist
• Set completeChapterCount = chapterBreakCount (not total chapter count)
• In-progress chapters are no longer treated as "new" chapters

Impact: Only complete chapters trigger detection and generation, preventing premature processing.

---

6. Lorebook Entry Format Inconsistency

Problem: generateChapterSummary() used Format B while regenerateChapter() used Format A, causing inconsistent lorebook entries.

Format A (old):

Chapter 5: Title Here
Type: chapter
Summary: Text here

Format B (standardized):

Chapter 5
Type: chapter
Title: Title Here
Summary: Text here

Solution:

• Standardized both functions to use Format B
• Fixed variable name typo in regenerateChapter() (messages → message)
• All lorebook entries now have consistent structure

Impact: All chapter summaries have uniform formatting regardless of how they were created.

---

7. Chapter 2 Title Extraction Bug

Problem: Chapter 2's lorebook entry would display Chapter 1's title instead of its own title (e.g., showing "Starship Hijinks" instead of "Bridge").

Root Cause: The isFirstChapter flag wasn't being cleared after chapter 1 was created via the auto-regeneration code path (regenerateChapter()), causing scanForPreviousChapter() to always return Chapter 1's text even when generating Chapter 2.

Solution:

• Added universal flag clearing logic to storeChapterFingerprint()
• When chapter 1's fingerprint is stored, set isFirstChapter = false
• This fix applies to both code paths: generateChapterSummary() and regenerateChapter()

Impact: All chapters now extract their correct titles regardless of which code path created them.

---

Configuration Modes

Version 1.5.2 properly implements three distinct behavior modes:

Mode 1: Full Manual (Both OFF)

• autoDetectOnGeneration: false
• autoRegenerate: false

Behavior: Nothing happens automatically. User must click "Check for Changes" button to detect and regenerate chapters manually.

Exception: First chapter of a brand new story is still auto-generated when user adds the first chapter break.

---

Mode 2: Semi-Automatic (Detect ON, Regenerate OFF)

• autoDetectOnGeneration: true
• autoRegenerate: false

Behavior:

• Script automatically detects changed/new chapters after each generation
• Shows notification modal with list of detected changes
• User manually clicks "Regenerate" or "Dismiss" for each chapter

---

Mode 3: Fully Automatic (Both ON)

• autoDetectOnGeneration: true
• autoRegenerate: true

Behavior:

• Script automatically detects changed/new chapters after each generation
• Automatically regenerates all detected changes in batch
• No user intervention required (fully hands-off)

---

Technical Improvements

Code Quality

• Removed obsolete multiChapterBatchInProgress flag
• Cleaned up redundant configuration checks
• Improved error handling and logging throughout
• Better separation of concerns between detection and generation

Timing & Logic

• More precise chapter completion detection
• Proper sequencing of automatic operations
• Better handling of edge cases (pasted stories, mixed scenarios)

Debugging

• Enhanced debug logging for flag state changes
• Better visibility into which code paths are executing
• Clear log messages for configuration mode detection

---

Migration Notes

No breaking changes - Existing configurations will continue to work. However, users may notice different behavior if they previously relied on the buggy automatic processing that bypassed configuration settings.

Recommended Action: Review your autoDetectOnGeneration and autoRegenerate settings to ensure they match your desired workflow (see Configuration Modes above).

---

Testing Coverage

All fixes have been tested with:

• ✅ Manual mode (both settings OFF)
• ✅ Semi-automatic mode (detect ON, regenerate OFF)
• ✅ Fully automatic mode (both ON)
• ✅ Pasted multi-chapter stories (10+ chapters)
• ✅ Writing chapters incrementally from scratch
• ✅ Mixed scenarios (existing chapters + adding new ones)
• ✅ Condensation triggering during various workflows
• ✅ Title extraction for all chapters
• ✅ Format consistency across all entry types

---

Known Limitations

None identified in v1.5.2 testing. All critical bugs from v1.5.1 and earlier have been resolved.

---

Credits

Special thanks to the testing that identified these edge cases and helped refine the three configuration modes into truly distinct, predictable behaviors.

---

Files Changed

• Automatic_Chapter_Summaries_v1.5.2.ts - Main script file with all fixes applied
• FEATURE_EXPLANATION_AUTO_VS_MANUAL.md - Documentation of the three behavior modes
• CHANGELOG_v1.5.2.md - This changelog

---

Upgrade Instructions

1. Replace your current script with Automatic_Chapter_Summaries_v1.5.2.ts
2. Review your configuration settings (autoDetectOnGeneration and autoRegenerate)
3. Read FEATURE_EXPLANATION_AUTO_VS_MANUAL.md to understand the three modes
4. Test with your current story to verify behavior matches expectations

---

Version 1.5.2 represents a major leap forward in reliability and user control. Enjoy the bug-free experience! 🎉

Release v1.5.1

Changelog - Version 1.5.1

Release Date: November 22, 2025

🎉 New Features

Condensed Ranges Management UI

• New Section in Main Panel: View all your condensed chapter ranges at a glance
• Range Details Display: Each range shows:
  • Chapter span (e.g., "Chapters 1-8")
  • Token count
  • Condensation date
  • Number of original summaries archived
• Quick Actions: "View Details" and "Uncondense" buttons for each range

View Condensed Range Details

• Comprehensive Modal: Shows complete information about any condensed range
  • Current condensed summary with full text
  • Metadata (token count, condensation date)
  • All archived original summaries with chapter numbers
  • Scrollable sections for easy navigation of long content

Uncondense Entire Range

• Preview Before Action: Confirmation modal shows:
  • Token impact calculation (+X tokens, +Y% increase)
  • List of chapters that will be restored
  • Option to proceed or cancel
• Smart Restoration: Deletes condensed entry and restores all original detailed summaries
• Fingerprint Updates: Automatically marks chapters as no longer condensed

Undo Last Uncondense Operation

• One-Click Undo: Reverse an uncondense operation if needed
• Automatic Backup: Stores condensed entry data before uncondensing
• UI Integration: Undo button appears when undo data is available
• Perfect for Testing: Facilitates safe experimentation and error recovery

Uncondense Single Chapter with Range Splitting

• Surgical Precision: Extract one chapter from a condensed range
• Smart Splitting:
  • Example: Uncondense Chapter 5 from "Chapters 3-7"
  • Result: "Chapters 3-4" (condensed) + Chapter 5 (detailed) + "Chapters 6-7" (condensed)
• Token Efficiency: Remaining chapters stay condensed to minimize token impact
• Single-Chapter Format: Uses "Chapter N" (not "Chapters N-N") for better readability
• Preview Modal: Shows split preview and token impact before proceeding
• Generation Limit Respect: Warns if operation would exceed 5-generation limit

🐛 Critical Bug Fixes

Fixed Overlapping Condensed Ranges Bug

Severity: Critical - Could corrupt lorebook state during automatic condensation

Symptoms:

• Multiple overlapping condensed ranges after auto-condensation
• Example: "Chapter 1", "Chapters 3-5", and "Chapters 1-8" all existing simultaneously
• Lorebook entries not properly cleaned up
• Storage array contained duplicate/overlapping ranges

Root Causes:

1. performNormalCondensation() was calling condenseSummaries() directly, bypassing overlap detection
2. condenseWithExpansion() was modifying the condensedRanges array during iteration
3. Function tried to delete same lorebook entries twice (overlap detection + deletion loop)
4. getChapterSummaryEntries() returns lorebook entries, not logical chapters, causing confusion

Solution:

• Complete rewrite of condenseWithExpansion() with clear 6-step process:
  1. Calculate new range boundaries from input chapters
  2. Find ALL overlapping ranges (before any modifications)
  3. Delete lorebook entries for overlapping ranges (with error handling)
  4. Collect summaries (from ranges + individuals, with duplicate prevention)
  5. Remove overlapping ranges from storage
  6. Create new condensed summary
• Changed performNormalCondensation() to use condenseWithExpansion()
• Added chaptersAdded Set to prevent duplicate chapters in summaries
• Each entry deleted exactly once with graceful error handling
• Clear separation: ranges provide originals, individuals provide themselves

Impact: Automatic condensation now properly cleans up old ranges and maintains clean lorebook state

Fixed Aggressive Condensation Duplicate Entries

Problem: When expanding condensed ranges during aggressive condensation, individual entries weren't being deleted, causing duplicates

Example Scenario:

• Had "Chapters 1-5" (condensed) + Chapters 6-9 (detailed)
• Aggressive condensation should create "Chapters 1-9"
• Chapter 9 wasn't deleted because only condensed entries were being removed
• Result: Both "Chapters 1-9" and old Chapter 9 entry existed

Solution:

• condenseWithExpansion() now tracks ALL entry IDs (condensed + individual)
• Deletes both condensed entries and individual entries before creating new range
• Created condenseSummariesWithoutDeletion() for pre-deleted entries
• Properly cleans up all old entries before creating new condensed range

Fixed Uncondense Single Chapter Errors

Problem: uncondenseSingleChapter() called condenseSummaries() which tried to delete non-existent entries

Details:

• Function was working with archived summaries (no actual lorebook entries)
• condenseSummaries() tried to delete entries by ID, causing errors
• Storage update: old range was filtered out but never saved, causing stale UI
• UI incorrectly showed old range instead of new split ranges

Solution:

• Changed to use condenseSummariesWithoutDeletion() for archived summaries
• Properly saves filtered condensedRanges array before creating new ranges
• UI now correctly displays only the new split ranges after operation

Fixed Duplicate Chapters in Condensed Summaries

Problem: When overlapping ranges existed, same chapter could be added to summariesToCondense twice

Example:

• "Chapter 1" (condensed 1-1) overlaps with new "Chapters 1-8" range
• Chapter 1 added from overlapping range's originalSummaries
• Chapter 1 also added from individual entries list
• Result: "Chapter 1" appears twice in archived summaries display

Solution:

• Added chaptersAdded Set to track processed chapter numbers
• Check before adding chapters from overlapping ranges
• Check before adding individual entries
• Prevents any chapter from being added more than once

🔧 Improvements

Single-Chapter Condensation Logic

• Token Efficiency Focus: When uncondensing creates single-chapter splits, they remain condensed
• Rationale: User wants ONE detailed chapter, others should stay token-efficient
• Format Consistency: Single-chapter entries use "Chapter N" everywhere:
  • Modal titles
  • Modal headers
  • Main panel display
  • Lorebook entry text
  • Type field: "Type: chapter" (not "Type: chapters")
• Automatic Condensation Safeguards:
  • Normal, aggressive, and emergency condensation all require minimum 2 chapters
  • Single-chapter condensed entries only created during manual uncondense splits
  • Prevents inappropriate automatic single-chapter condensation

Enhanced UI Feedback

• Status Panel Improvements:
  • Accurate detailed/condensed entry counts using storage data (not lorebook scan)
  • Clear display of which chapters are in condensed ranges
  • Better visual separation between sections
• Modal Consistency: All modals use consistent formatting for singular vs. plural chapters
• Generation Limit Tracking: Counter properly increments for all condensation operations

Code Architecture Improvements

• Separation of Concerns:
  • condenseSummaries() - normal condensation with deletion
  • condenseSummariesWithoutDeletion() - condensation for pre-deleted entries
  • condenseWithExpansion() - condensation with overlap handling
• Better Error Handling: Graceful handling of missing/already-deleted entries
• Cleaner Logic Flow: Step-by-step process easier to understand and debug

📋 Technical Details

New Functions

• buildCondensedRangesUI() - Builds UI section for condensed ranges display
• showCondensedRangeDetails() - Opens detailed modal for a condensed range
• confirmUncondenseRange() - Shows preview modal before uncondensing entire range
• uncondenseEntireRange() - Deletes condensed entry and restores all originals
• undoLastUncondense() - Reverses last uncondense operation
• confirmUncondenseSingleChapter() - Shows preview for single chapter uncondense
• uncondenseSingleChapter() - Entry point with generation limit checking
• performUncondenseSingleChapter() - Internal function performing split operation
• condenseSummariesWithoutDeletion() - Condenses pre-deleted entries

Modified Functions

• condenseWithExpansion() - Complete rewrite with 6-step clear process
• performNormalCondensation() - Now uses condenseWithExpansion()
• updateStatusPanel() - Fixed count calculations using storage data
• getChapterSummaryEntries() - Enhanced logging and validation

Storage Keys

• lastUncondenseData - Stores backup for undo functionality
• condensedRanges - Enhanced with proper overlap management

🎮 Usage Tips

1. View Your Condensed Ranges: Check the "📦 Condensed Ranges" section in the main panel
2. Need Details?: Click "View Details" on any range to see the full condensed summary and all archived originals
3. Want Full Detail Back?: Click "Uncondense" to restore all original detailed summaries (with token impact preview)
4. Made a Mistake?: Use the "Undo Last Uncondense" button (appears after uncondense operations)
5. Extract One Chapter: In the details modal, use "Uncondense Single Chapter" to get just one chapter back in detail while keeping others condensed
6. Watch Your Tokens: All operations show token impact before proceeding

⚠️ Breaking Changes

None - All changes are backward compatible with v1.5.0 data structures.

🔮 Future Enhancements

Potential improvements for future versions:

• Batch uncondense multiple ranges

---

Upgrade Note: This version requires no data migration. Your existing condensed ranges, fingerprints, and backups will work seamlessly with the new features.

v1.5.0

What's Changed

• Chapter Summarize v1.5.0: Enhanced backup browser by @LaneLearns in #2

Full Changelog: v1.4.1...v1.5.0

v1.4.1

What's Changed

• Release of v1.4.1 - Automatic Chapter Summary Lorebook Entries via co… by @LaneLearns in #1

New Contributors

• @LaneLearns made their first contribution in #1

Full Changelog: https://github.com/LaneLearns/AutomaticChapterSummaries/commits/v1.4.1