- reconcileIndexedDbProbeFailureState now handles persist-mismatch:indexeddb-* reasons, ensuring bounded retries and transition to BLOCKED instead of hanging forever in LOADING
- expose clearCachedIndexedDbSnapshot, clearAllCachedIndexedDbSnapshots, closeBmeDb, closeAllBmeDbs, syncGraphLoadFromLiveContext in runtime context for UI actions
- onDeleteCurrentIdbController / onDeleteAllIdbController now clear in-memory caches, close DB handles, and force a fresh graph load sync after deletion
- humanize persist-mismatch:indexeddb-behind-commit-marker panel messages into user-friendly Chinese text
- add regression test for empty IndexedDB + accepted commit marker mismatch fallback-to-blocked scenario
- Add hasContiguousJournalCoverageThroughFloor to check if current
batchJournal contiguously covers back through the old retained floor
- Add reconcileManualBackupBatchJournalCoverage that clears stale
coverage when actual journals have bridged the gap
- Self-clean during normalizeGraphRuntimeState (load-time) and
appendBatchJournal (write-time)
- findJournalRecoveryPoint now uses reconciled coverage so stale
manual backup floors no longer block valid reverse-journal recovery
- Add regression tests for gap-not-yet-bridged and bridged scenarios
- prompt-builder.js: add RECALL_TARGET_CONTENT_HEADER, update splitSectionedTranscriptPayloadMessage to recognize recall-specific target header
- retriever.js: add buildRecallSectionedTranscript helper, format recentMessages as sectioned transcript with context-review and recall-target headers for prompt building while keeping flat string[] for ranking
- p0-regressions.mjs: add testRecallUsesSectionedPromptMessagesForContextAndTarget regression asserting two system messages with correct transcriptSection and headers
Root cause: formatPromptMessageTranscript in prompt-builder.js ignored
isContextOnly, so context review and extraction target sections were
flattened into plain transcript even though the flag was correctly set
in intermediate layers. Additionally, userPromptSections (which
contained the dividers) was only a fallback that never reached the
final prompt when block-based profiles had user blocks.
Fix:
- getPromptMessageLikeDescriptor now preserves isContextOnly flag
- formatPromptMessageTranscript now inserts context/target section
dividers when messages carry isContextOnly, ensuring the final
LLM prompt always shows the distinction regardless of which
rendering path (recentMessages, chatMessages, dialogueText) is used
Regression tests:
- prompt-builder-mixed-transcript: verify recentMessages block content
includes context review and extraction target dividers
- extractor-phase3-layered-context: end-to-end test proving default
extract profile + default structured mode produces final promptMessages
with context/target section dividers
- Add extractActionMode setting (default: pending) to settings-defaults.js
- Panel refresh restores dropdown from settings.extractActionMode
- Dropdown change event persists selection via _patchSettings
- Extract button reads settings as fallback when DOM value missing
- Add default value assertion in tests/default-settings.mjs
