Backfill on a 1:1 chat with ~96 messages finds unextracted messages and calculates batch count, then immediately contradicts itself and exits without processing anything. This happens consistently on this specific chat — other 1:1 chats with similar message counts backfill normally using the same Connection Profile and settings.
Console output (relevant lines):
[OpenVault] Backfill state: 0 processed messages tracked, 0 memories stored
[OpenVault] Backfill check: 96 unextracted messages available, 6 complete batches remaining
[OpenVault] Backfill: No more complete batches available
[OpenVault] Backfill Phase 1 complete. Running final Phase 2 synthesis...
[OpenVault] runPhase2Enrichment: No memories to enrich
[OpenVault] Extension enabled - event listeners registered
[OpenVault] Backfill complete
[OpenVault] Stats: 0 memories, 0 embeddings, 0 characters
This runs twice in a row with identical output — 96 messages found, 6 batches calculated, then immediately "no more complete batches."
What I expected: Backfill should process the 6 batches it detected.
What happened: Zero batches processed, zero memories extracted.
Context:
- Other 1:1 chats with similar length backfill fine (e.g., a Jin/Ying chat processed 7 batches, 14 memories, 18 total with reflections)
- Same Connection Profile used for both
- SillyTavern version: [fill in your version]
- OpenVault version: [check your manifest.json or the version shown in the panel]
- Chat is 1:1, not a group chat
- Extension was freshly installed on this chat (no prior OpenVault data)
Additional note: Graph extraction and community summarization also fail consistently due to a separate issue — the Connection Profile routes through KoboldCpp Text Completion API, which returns JSON in the reasoning field with empty content. Event extraction works because the simpler prompt sometimes gets JSON into content, but graph/community fail 100%. I'm switching to a Chat Completion profile to fix that separately, but the backfill batch contradiction happens before any LLM call is made, so it's a different issue.
Debug export:
{
"openvault_debug_export": true,
"exportedAt": "2026-04-24T19:01:46.538Z",
"lastRetrieval": null,
"scoring": null,
"state": {
"memories": { "total": 0, "byType": {}, "byImportance": {}, "averageImportance": 0 },
"characterStates": {},
"graph": { "summary": { "nodeCount": 0, "edgeCount": 0, "typeBreakdown": {}, "topEntitiesByMentions": [] } },
"communities": { "count": 0, "details": {} }
},
"settings": {
"extractionProfile": "15d00181-c394-4cdd-86ca-3ae9c0b515a9",
"backupProfile": "15d00181-c394-4cdd-86ca-3ae9c0b515a9",
"extractionTokenBudget": 4000,
"extractionRearviewTokens": 4000,
"retrievalFinalTokens": 2000,
"autoHideEnabled": false,
"maxConcurrency": 2
},
"runtime": {
"extractionProgress": { "processed": 0, "chatLength": 101 }
}
}
Backfill on a 1:1 chat with ~96 messages finds unextracted messages and calculates batch count, then immediately contradicts itself and exits without processing anything. This happens consistently on this specific chat — other 1:1 chats with similar message counts backfill normally using the same Connection Profile and settings.
Console output (relevant lines):
This runs twice in a row with identical output — 96 messages found, 6 batches calculated, then immediately "no more complete batches."
What I expected: Backfill should process the 6 batches it detected.
What happened: Zero batches processed, zero memories extracted.
Context:
Additional note: Graph extraction and community summarization also fail consistently due to a separate issue — the Connection Profile routes through KoboldCpp Text Completion API, which returns JSON in the
reasoningfield with emptycontent. Event extraction works because the simpler prompt sometimes gets JSON into content, but graph/community fail 100%. I'm switching to a Chat Completion profile to fix that separately, but the backfill batch contradiction happens before any LLM call is made, so it's a different issue.Debug export:
{ "openvault_debug_export": true, "exportedAt": "2026-04-24T19:01:46.538Z", "lastRetrieval": null, "scoring": null, "state": { "memories": { "total": 0, "byType": {}, "byImportance": {}, "averageImportance": 0 }, "characterStates": {}, "graph": { "summary": { "nodeCount": 0, "edgeCount": 0, "typeBreakdown": {}, "topEntitiesByMentions": [] } }, "communities": { "count": 0, "details": {} } }, "settings": { "extractionProfile": "15d00181-c394-4cdd-86ca-3ae9c0b515a9", "backupProfile": "15d00181-c394-4cdd-86ca-3ae9c0b515a9", "extractionTokenBudget": 4000, "extractionRearviewTokens": 4000, "retrievalFinalTokens": 2000, "autoHideEnabled": false, "maxConcurrency": 2 }, "runtime": { "extractionProgress": { "processed": 0, "chatLength": 101 } } }