test Supermemory with native Gemini

servers/unraid/supermemory-poc/README.md

@@ -10,10 +10,10 @@ production Hindsight service or change the default Hermes profile.
   cache live under `/mnt/user/appdata/supermemory-poc/data`.
 - A dedicated address on Docker `br0`; port `6767` is not published on the
   unRaid host address.
-- Remote OpenAI-compatible embeddings through HK LiteLLM using
-  `google/gemini-embedding-2` at 1024 dimensions.
-- Extraction, summarization, and other intelligent steps use
-  `gemini-3.5-flash-lite` through the same LiteLLM endpoint.
+- Native Gemini embeddings using `gemini-embedding-2` at 768 dimensions.
+- Extraction, summarization, and other intelligent steps use Supermemory's
+  native Gemini provider default. This cleanly removes the LiteLLM and
+  OpenAI-compatible translation layer from the diagnostic path.
 - Ingest concurrency remains 1 so the first POC emphasizes correctness and
   observable cost over throughput.
 - Resource ceiling: 4 CPUs and 8 GiB RAM. Supermemory Local keeps its corpus
@@ -26,11 +26,10 @@ directory and re-ingest when comparing a different embedding plan.
 
 1. Confirm the chosen `SUPERMEMORY_IPV4_ADDRESS` is absent from both Arcane's
    `br0` attachments and the LAN neighbor/DHCP tables.
-2. Add `SUPERMEMORY_IPV4_ADDRESS`, `OPENAI_API_KEY`,
-   `OPENAI_BASE_URL=https://api.opfo.xyz/v1`, and
-   `OPENAI_MODEL=gemini-3.5-flash-lite` to this Project's Arcane environment.
-   `OPENAI_API_KEY` is the HK LiteLLM bearer key. Do not put it in Git,
-   Compose, activity notes, or chat.
+2. Add `SUPERMEMORY_IPV4_ADDRESS` and `GEMINI_API_KEY` to this Project's
+   Arcane environment. Do not put the key in Git, Compose, activity notes, or
+   chat. Legacy `OPENAI_*` values may remain stored in Arcane for rollback,
+   but this Compose file does not inject them into the container.
 3. Confirm `/mnt/user/appdata` has room for the 298 MiB server binary, the
    multilingual model cache, data, and rollback copy.
 4. Confirm no other build, image pull, backup, or migration is active on the
@@ -49,7 +48,7 @@ image build is required.
 
 The server creates its client bearer token at:
 
-`/mnt/user/appdata/supermemory-poc/data-google-gemini-embedding-2-1024/api-key`
+`/mnt/user/appdata/supermemory-poc/data-gemini-embedding-2-768/api-key`
 
 Read it through an authorized unRaid/Arcane console without printing it into
 logs. Save it only as `SUPERMEMORY_API_KEY` in the `supermemory-lab` Hermes
@@ -83,24 +82,23 @@ used by Hermes.
 
 ### Embedding plan and smoke gate
 
-The earlier local `Xenova/bge-m3` experiment is preserved under the old `data`
-directory because embedding plans cannot be changed in place. This deployment
-uses the fresh `data-google-gemini-embedding-2-1024` directory. The one-shot
-`supermemory-smoke` service verifies the locked embedding plan, performs a
-Chinese hybrid-search round trip, checks provenance metadata, and exercises
-the profile and conversation endpoints. It writes a sentinel after success so
-ordinary redeploys do not add duplicate smoke documents. Do not import sampled
+The earlier local `Xenova/bge-m3` and LiteLLM-backed embedding experiments are
+preserved under their old data directories because embedding plans cannot be
+changed in place. This deployment uses the fresh
+`data-gemini-embedding-2-768` directory. The one-shot `supermemory-smoke`
+service verifies the locked embedding plan, tracks the exact document ID to a
+terminal state, separately checks v3 search, v4 document search, and Chinese
+hybrid search, checks provenance metadata, and exercises the profile and
+conversation endpoints. It writes a sentinel after success so ordinary
+redeploys do not add duplicate smoke documents. Do not import sampled
 Hindsight or document material until this service exits successfully.
 
-On 2026-09-08, the remote plan was accepted and locked as
-`openai / google/gemini-embedding-2 / 1024`; the LiteLLM embedding and
-`gemini-3.5-flash-lite` chat endpoints also passed independently. Supermemory
-listed the smoke document and reported a completed document in the test
-container, but exact searches for its unique marker returned no hit through
-either `/v4/search` or `/v3/search` (`supermemory-smoke` exit 44). Treat this
-as a self-hosted indexing/retrieval blocker. Do not import sampled material or
-enable the Hermes lab profile until a newer server build or upstream fix makes
-the same smoke pass.
+On 2026-09-08, the previous remote plan was accepted and locked as
+`openai / google/gemini-embedding-2 / 1024`; the LiteLLM embedding and chat
+endpoints also passed independently. Supermemory listed a completed smoke
+document, but both v4 and v3 searches returned no hit. The old smoke mixed
+list-wide status checks and did not track the newly created document, so this
+native Gemini run uses a new store and a stricter document-specific gate.
 
 ## Evaluation guardrails