From 2d1503921d274d0183664d282b8c762849c057b7 Mon Sep 17 00:00:00 2001
From: happybell80 <happybell80@naver.com>
Date: Sat, 21 Mar 2026 11:01:50 +0900
Subject: [PATCH] =?UTF-8?q?docs:=20=EC=BD=9C=EB=93=9C=EB=A9=94=EC=9D=BC=20?=
 =?UTF-8?q?=EB=A6=AC=EC=8A=A4=ED=8A=B8=20=ED=98=84=ED=99=A9=20=EA=B8=B0?=
 =?UTF-8?q?=EB=A1=9D,=20OCR=20fallback=20=EC=9E=90=EB=8F=99=20=EB=8F=99?=
 =?UTF-8?q?=EC=9E=91=20=EB=AA=85=EC=8B=9C?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>
---
 skills/skill-rag-file/SKILL.md | 1 +
 skills/skill-slack/SKILL.md    | 7 +++++++
 2 files changed, 8 insertions(+)

diff --git a/skills/skill-rag-file/SKILL.md b/skills/skill-rag-file/SKILL.md
index e462fa1..67d762f 100644
--- a/skills/skill-rag-file/SKILL.md
+++ b/skills/skill-rag-file/SKILL.md
@@ -35,6 +35,7 @@ RAG(Retrieval-Augmented Generation) 파일 처리 스킬. 포트 8508.
 - 모델: `OCR_MODEL=gemini-2.0-flash` (runtime.env SSOT)
 - 용도: 이미지 기반 PDF 텍스트 추출 (text_length 부족 시 선별 적용)
 - 벤치마크: 3.4초/페이지, 2.3초/MB — 속도+품질 최적 ([리서치](../../journey/research/rag/260320_OCR_모델_벤치마크_리서치.md))
+- IR덱 평가 시 텍스트 추출 실패 → OCR fallback 자동 동작 (`ir_deck_workflow.py`)
 
 ## 환경변수
 
diff --git a/skills/skill-slack/SKILL.md b/skills/skill-slack/SKILL.md
index 88bfb9e..205c275 100644
--- a/skills/skill-slack/SKILL.md
+++ b/skills/skill-slack/SKILL.md
@@ -76,6 +76,13 @@ Slack 통합 스킬. 포트 8502.
 | `ir_deck_page_evaluations` | 장표별 page_number, score, strengths[], weaknesses[], page_topic |
 | `coldmail_memory_email` | evaluation_id, document_id, ir_deck_path, robeing_evaluation |
 
+### 현황 (260320)
+
+- 슬랙 리스트: 57건 (고유 53개 회사), 전건 평가 완료
+- DB(`coldmail_memory_email`): 60건 (테스트 3건 포함), 59건 평가+문서 연결
+- 중복 기업: 에듀온(3), 누리광장(2), 담을거야(2), NAVER(2)
+- OCR fallback 적용: 이미지 기반 PDF도 Gemini 2.0 Flash로 평가 가능
+
 ### 확인 방법
 
 ```bash