📄 中文摘要
为了在云端运行自己的 AI 代理，而不是在个人电脑或其他人托管的仪表板上，选择了完全自托管的方式。通过在 AWS 免费层 EC2 实例上部署 OpenClaw，并使用 Tailscale 进行安全保护，成功将其与 Telegram 通过 Gemini API 连接。该过程详细说明了具体的设置步骤，确保操作简单有效。没有多余的内容，直接提供可行的解决方案。
我在 AWS 上自托管 OpenClaw，花费 $0 — 无开放端口，无 SaaS，无妥协（使用 TailScale）
📄 English Summary
🦞 I Self-Hosted OpenClaw on AWS for $0 — No Open Ports, No SaaS, No Compromise (Using TailScale)
The goal was to run a personal AI agent in the cloud, fully self-hosted and under personal control, rather than on a laptop or someone else's hosted dashboard. OpenClaw was deployed on a free-tier AWS EC2 instance, secured with Tailscale, and connected to Telegram via the Gemini API. The setup process is outlined in detail, providing a straightforward and effective solution without unnecessary fluff.