diff --git a/README.ja.md b/README.ja.md
index 643761d..ecce707 100644
--- a/README.ja.md
+++ b/README.ja.md
@@ -93,7 +93,7 @@ robocoが常時運用している**GPUスポット確保可能性(プレイス
 cdkのコンテキストとscripts/*.shを直接扱う代わりに、統合CLIを使うことができる:
 
 ```bash
-npm install -g tkf   # tkfコマンドをインストール
+npm install -g @serithemage/tkf   # tkfコマンドをインストール
 # もしくはソースからインストール:
 # npm install && npm run build && npm link
 tkf model list                              # 検証済みモデルカタログ
diff --git a/README.ko.md b/README.ko.md
index 5921949..48b0c79 100644
--- a/README.ko.md
+++ b/README.ko.md
@@ -87,7 +87,7 @@ roboco가 상시 운영하는 **GPU 스팟 확보 가능성(배치점수) × 가
 cdk 컨텍스트와 scripts/*.sh를 직접 다루는 대신 통합 CLI를 쓸 수 있다:
 
 ```bash
-npm install -g tkf   # tkf 명령 설치
+npm install -g @serithemage/tkf   # tkf 명령 설치
 # 또는 소스에서 설치:
 # npm install && npm run build && npm link
 tkf model list                              # 검증된 모델 카탈로그
diff --git a/README.md b/README.md
index 4f6d62d..0a197ee 100644
--- a/README.md
+++ b/README.md
@@ -103,7 +103,7 @@ Instead of working with cdk context flags and scripts/*.sh directly, you can use
 unified CLI:
 
 ```bash
-npm install -g tkf   # install the tkf command
+npm install -g @serithemage/tkf   # install the tkf command
 # or, to install from source:
 # npm install && npm run build && npm link
 tkf model list                              # verified model catalog
diff --git a/package-lock.json b/package-lock.json
index c52d80b..ac01637 100644
--- a/package-lock.json
+++ b/package-lock.json
@@ -1,11 +1,11 @@
 {
-  "name": "tkf",
+  "name": "@serithemage/tkf",
   "version": "1.0.0",
   "lockfileVersion": 3,
   "requires": true,
   "packages": {
     "": {
-      "name": "tkf",
+      "name": "@serithemage/tkf",
       "version": "1.0.0",
       "license": "MIT",
       "dependencies": {
diff --git a/package.json b/package.json
index 2108e02..0ea7d66 100644
--- a/package.json
+++ b/package.json
@@ -1,5 +1,5 @@
 {
-  "name": "tkf",
+  "name": "@serithemage/tkf",
   "version": "1.0.0",
   "description": "Private vibe-coding LLM in your own AWS account — open-weight models on 100% spot instances with placement-score intelligence",
   "keywords": [