diff --git a/.github/workflows/openie-fixture.yml b/.github/workflows/openie-fixture.yml
index 4e570a7..9a47b0e 100644
--- a/.github/workflows/openie-fixture.yml
+++ b/.github/workflows/openie-fixture.yml
@@ -29,7 +29,7 @@ jobs:
           restore-keys: uv-${{ runner.os }}-
 
       - name: Install dependencies
-        run: uv sync --extra dev --extra sqlite
+        run: uv sync --extra dev --extra sqlite --extra embedding
 
       - name: Create synthetic OpenIE fixture
         run: |
diff --git a/docs/REPORT-memory-operating-layer-eval.md b/docs/REPORT-memory-operating-layer-eval.md
index 99c818a..2349154 100644
--- a/docs/REPORT-memory-operating-layer-eval.md
+++ b/docs/REPORT-memory-operating-layer-eval.md
@@ -1844,7 +1844,9 @@ per-chunk fallback을 유지한다.
   `openie_mz_poc.py --skip-openie`를 scheduled/manual 실행한다. GitHub-hosted runner에서
   ingest, scoring, gate semantics, revertibility가 계속 연결되어 있는지 감시하고,
   `/tmp/openie_fixture_results.json` artifact를 남긴다. 로컬 workflow-equivalent smoke는
-  fixture query `R@1=100%`, `R@5=100%`, revertibility `PASS`로 완료됐다.
+  fixture query `R@1=100%`, `R@5=100%`, revertibility `PASS`로 완료됐다. 첫 remote dispatch는
+  fixture smoke는 통과했지만 contract test preflight에서 `aiohttp`가 없어 실패했고,
+  workflow dependency를 `--extra embedding`까지 설치하도록 보정했다.
 - PR #15의 300-edge SQLite micro-benchmark는 old per-edge write
   `109,962.04ms` 대비 batch write `195.85ms`로 `561.45x` 빨랐다.
 - PR #17의 100-chunk repeated-entity micro-benchmark는 backend `get_node()` `1`,