TencentDB-Agent-Memory · Issues· 786 open

[Docs] Codex integration silently requires an upstream that implements the Responses API; MemoryProxy forwards /v1/responses without translation

[shepherding] Pi five-role skill capture pipeline + `/tdai-memory-sync-skills`

[Bug] L1 extraction failures advance the L0→L1 cursor and silently drop conversations

[Bug] Pi sessions are missing from Skill and Memory Bridge L1 prefix fallback

Pi provider sends developer role to strict OpenAI-compatible upstreams

[Bug] Pi profile splits sections inside nested XML blocks — memory block injected twice

[Feature] No way to control reasoning/thinking in the standalone LLM path (truncated responses become empty extractions)

[Bug] L1 extraction can attribute a third party's name to the user (no name-source rule in the prompt)

[Feature] Per-layer (L1/L2/L3) model overrides and multi-provider LLM config for standalone mode

[shepherding] `/tdai-memory-forget` — interactive redacted-preview memory deletion

[shepherding] Pi `/tree` branch isolation via `tdai-memory/branch@1` marker

[shepherding] Durable outbox for Pi adapter: atomic leases + at-least-once + dead-letter isolation

[Bug][Hermes] README's MEMORY_TENCENTDB_LLM_* vars never reach the Gateway — L1 extraction fails silently while health stays green

[Bug] Wiki asset metadata not persisted after ingestion pipeline completes

recall.scoreThreshold applied across incommensurable score scales (BM25/cosine 0-1 vs RRF ~0.03)