修改发布脚本，微调内容

Author: fooying

.github/workflows/deploy.yml

@@ -49,10 +49,18 @@ jobs:
             --minify \
             -b 'https://sec.cafe/handbook/'
       - name: upload code
-        uses: garygrossgarten/github-action-scp@release
+        uses: appleboy/scp-action@v0.1.7
         with:
-          local: public/.
-          remote: /data/www/handbook.sec.cafe/
+          source: public/*.*
+          target: /tmp/
+          host: ${{ secrets.DEPLOY_HOST }}
+          port: ${{secrets.DEPLOY_HOST_PORT}}
+          username: ${{ secrets.DEPLOY_HOST_USER}}
+          key: ${{ secrets.DEPLOY_PRIVATE_KEY}}
+      - name: deploy ui
+        uses: garygrossgarten/github-action-ssh@release
+        with:
+          command: cd /data/www/handbook.sec.cafe/;rm -rf *;mv /tmp/public/* .;rm -rf /tmp/public
           host: ${{ secrets.DEPLOY_HOST }}
           port: ${{secrets.DEPLOY_HOST_PORT}}
           username: ${{ secrets.DEPLOY_HOST_USER}}

content/security_research/ai_security/llm_security/attack.md

@@ -23,6 +23,8 @@ keywords:
 - [金融行业的人工智能安全风险研究](https://www.freebuf.com/articles/paper/378893.html)
 - [大语言模型 (Large Language Model，简称LLM) 安全之攻击面初探](https://mp.weixin.qq.com/s/GMmbyMvqKxm0z1HPQeTgag)
 - [实战解读：Llama 3 安全性对抗分析](https://mp.weixin.qq.com/s/3OqoJIqSGLIMa1dVzXXFkQ)
+- [实战解读：Llama Guard 3 & Prompt Guard](https://mp.weixin.qq.com/s/jgwn_o0EUmvtNUhGswCDng)
+
 
 ## 测试框架与方法
 - [针对大语言模型的通用对抗性攻击](https://www.freebuf.com/articles/paper/375497.html)

content/security_research/ai_security/llm_security/defense.md

@@ -32,8 +32,6 @@ keywords:
 - [vivo对生成式人工智能的安全思考和实践](#)
 - [OpenAI安全系统负责人长文梳理：大模型的对抗攻击与防御](https://mp.weixin.qq.com/s/t87IOi6r4N-c-StI9CPy_A)
 - [Meta开源大模型的安全实践](https://mp.weixin.qq.com/s/-mHiEfImfZBgotDgRnKU5w)
-- [实战解读：Llama 3 安全性对抗分析](https://mp.weixin.qq.com/s/3OqoJIqSGLIMa1dVzXXFkQ)
-- [实战解读：Llama Guard 3 & Prompt Guard](https://mp.weixin.qq.com/s/jgwn_o0EUmvtNUhGswCDng)
 
 ## 工具
 - [Garak](https://docs.garak.ai/garak/) 一个用于大语言模型 (LLM) 漏洞扫描的 Python 包