I Taught My Coding Agent to Run My Server With One Text File. Here’s How to Write One for Yours.

说明与用途

作者创建了一个单一的文本文件，教会 AI 代理如何运行和管理服务器，消除了手动粘贴指令的需要。该文件定义了技能和斜杠命令，自动执行部署、状态检查和其他任务，减少错误并加速设置。

问题

Agents need a concise, machine‑readable description of a server’s deployment and operation details, but users had to paste the same paragraph into every new chat, leading to errors and inefficiency.

结果

After implementing the skill file approach, the author no longer needs to paste deployment instructions manually; the agent can sign up, set up, and operate the server from a single link, improving reliability and reducing setup time.

工作流

• 1. Create a skill file (SKILL.md) with a header and detailed instructions for agentic‑hosting.
• 2. Publish the skill to a raw GitHub URL (e.g., https://raw.githubusercontent.com/dennisonbertram/agentic-hosting/main/skills/agentic-hosting/SKILL.md).
• 3. Create an index file (llms.txt) that points to the skill and other docs.
• 4. Install the skill on the agent via a single shell command or by downloading the file into ~/.claude/skills/agentic-hosting.
• 5. Export environment variables AH_URL and AH_KEY with the server address and API key.
• 6. Use natural language or slash commands (e.g., /ah-deploy) to perform deployment, status checks, database addition, etc. The agent reads the skill and executes curl calls automatically.
• 7. Adjust slash command headers to use correct argument numbering ($0, $1) or $ARGUMENTS to avoid bugs.

证据

I had a ritual. Every time I opened a new chat with my coding agent, I pasted in the same paragraph: here's my server's address, here's how you deploy, the build field is called this and not that, don't do the other thing.

来源

That paragraph is now a file. It sits in a folder on my laptop, the agent reads it when the conversation turns to my server, and I haven't pasted anything in months.

来源

The six slash commands are shortcuts for the jobs I do most: /ah-deploy, /ah-status, /ah-db, /ah-logs, /ah-register, /ah-snapshot.

来源

Slash commands can pull in what you typed after the command. My command files were written when $1 meant the first argument. In current Claude Code, numbering starts at zero: $0 is the first argument and $1 is the second.

来源