diff --git a/troubleshooting/251014_claude_coldmail_filter_tokenization_issue.md b/troubleshooting/251014_claude_coldmail_filter_tokenization_issue.md
index 7fa039a..c26a6e2 100644
--- a/troubleshooting/251014_claude_coldmail_filter_tokenization_issue.md
+++ b/troubleshooting/251014_claude_coldmail_filter_tokenization_issue.md
@@ -46,7 +46,7 @@
 - calculate_similarity(): cosine similarity
 - is_coldmail_by_embedding(): threshold 0.6
 
-**DB**: coldmail_embedding_clusters 테이블 (pgvector extension 필요)
+**DB**: coldmail_embedding_clusters 테이블 (pgvector extension, VECTOR(384) 필요)
 
 ### Phase 2: Gemini LLM 분류기 (2h)
 **신규**: coldmail_llm_classifier.py
@@ -80,7 +80,7 @@
 ## 구현 완료
 
 ### 테스트 스크립트 (커밋 7c5b033)
-**파일**: rb8001/tests/test_coldmail_filter.py
+**파일**: rb8001/tests/test_coldmail_briefing.py
 - test_hybrid_coldmail_filter():92-134 추가
 - --test hybrid 옵션 추가