大家好，我是：
- 农村程序员（2020年回农村，至今未出村）
- 独立开发者（致力于开发 1000 个小产品）
- 技术观察者（分享编程技术，行业资讯等）
- 前端之虎陈随易（自封的）
- 我的官网是：https://yicode.tech
给大家分享一个 AI 运维服务器，安全又稳定的方案。
这个方案就是 Finch Agent 工具 + 远程主机小程序，Finch Agent 工具下载地址：https://finchwork.app/zh
如图，我在聊天框发送 显示当前主机的基本信息，就能显示远程主机的基本配置。
我们可以随时切换主机。
切换主机后，再次发送 显示当前主机的基本信息，显示的就是选定主机的配置了。
我们可以重启服务，诊断问题，修改文件等等，再也不用手动敲命令，查资料。
我们的每一个操作记录，操作命令，都会在对应主机的操作中记录下来。
服务器可是重要的资产，万一 AI 乱来怎么办？误操作怎么办？
基于安全问题，敏感操作问题，我们可以自定义白名单命令，黑名单命令，用 glob 语法匹配。
同时，提供了 4 种确认策略，可以每个命令确认，也可以让读取操作自动放行，其他操作询问。
同时，为了防止绕过，远程主机小程序 内置了强制性的安全规则，比如严禁删除 / 根目录，以及 /var、/home、/etc 等一级目录，严禁写裸盘，fork，关机重启等。
敏感操作，必须自己登录主机，手动操作，避免 AI 提供命令，自己顺手确认，造成不可挽回的问题。
如果大家感兴趣，可以下载 Finch Agent 工具，下载地址：https://finchwork.app/zh，找到 远程主机小程序，安装并启用。
这里值得说道的是，通过这个小程序，我们的服务器 IP 地址、账号、密码等，都完全不用暴露在 AI 的聊天窗口和会话里。
比如我们提供服务器账号，密码，让 AI 连接，操作，还没有任何安全防护要好得多。
远程主机小程序启用后，打开侧边栏，添加主机，最后在聊天窗口下方，选择主机，就能对其进行各种操作了。
当然，百密一疏，远程主机小程序 刚发布，还有很多问题需要大家的反馈、完善、优化。
本文由 AI图鉴自动抓取并翻译，版权归原作者所有；如需查看原文请点下方链接。
Hello, I'm:
- Rural programmer (back to the countryside in 2020, not yet out of the village)
- Independent developers.
- Technical observers (sharing of programming techniques, industry information, etc.)
- The Tiger of the Front.
- My network is: https://yicode.tech
Share with you an AI traffic server, secure and stable.
This scheme is the FinchAgent Tool + Remote Host Applet, which is a downloading address for the FinchAgent Tool: https://finchwork.app/zh
As illustrated, I send basic information on the current host in the chat box to show the basic configuration of the remote host.
We can switch the mainframe anytime.
Toggle the host, send again the basic information about the current host, which is the configuration of the selected host.
We can restart services, diagnose problems, modify documents, etc., and never use manual taps to check information.
Each of our operational records, operational orders, will be recorded in the corresponding mainframe operation.
Servers are important assets. What if AI messes up? What about the mistake?
On the basis of security concerns, sensitive operations, we can customize white list commands, blacklist commands, and match them with glob syntax.
At the same time, 4 confirmation strategies are provided, which can be confirmed by each command, or read operations can be automatically released and other operations asked.
At the same time, in order to prevent circumvention, the remote host applets incorporate mandatory security rules, such as a strict ban on removing / root directories, and a first-level catalogue such as /var, /home, /etc, a strict ban on writing nudity disks, fork, shutdown, etc.
Sensitive operations, which must log in to the host themselves, operate manually, avoid AI providing orders and confirm them by itself, creating irreparable problems.
If you are interested, you can download the FinchAgent tool, download the address: https://finchwork.app/zh, find the remote host applet, install it and enable it.
It's worth saying that through this little program, our server IP addresses, accounts, passwords, etc., are not exposed at all to AI chat windows and sessions.
For example, we provide server accounts, passwords, connect AI, operate without any security protection.
When the remote host applet is enabled, the sidebar is opened, the host is added, and finally the host is selected below the chat window.
And, of course, it's a little short, remote host applet, just publish, and there's a lot of questions that need feedback, refinement and optimization.
Auto-fetched and translated by AI Atlas. All rights belong to the original author.