OpenClaw AWS Deploy — OpenClaw AWS 部署

OpenClaw AWS 部署 技能 Quick 启动 (Minimal 部署ment ~$30/mo) Prerequisites AWS 凭证s — any of these methods: --性能分析 flag (named AWS 命令行工具 性能分析) .env.aws file in workspace root or 技能 directory (optional): AWS_访问_KEY_ID=... AWS_SECRET_访问_KEY=... AWS_DEFAULT_REGION=us-east-1

Existing 环境 variables, AWS SSO 会话, or IAM 角色 .env.starfish in workspace root (recommended) or 技能 directory: TELEGRAM_机器人_令牌=... # from @机器人Father (required) TELEGRAM_USER_ID=... # your Telegram user ID (optional, enables auto-应用rove pAIring) GEMINI_API_KEY=... # from AIstudio.google.com (optional, for Gemini 模型s)

aws 命令行工具 安装ed OR Docker for sandboxed 访问 jq, openssl avAIlable One-Shot 部署 # From the 技能 directory: ./scripts/部署_minimal.sh --name starfish --region us-east-1 \ --env-dir /path/to/workspace

# Or with 清理up of previous 部署ment first: ./scripts/部署_minimal.sh --name starfish --region us-east-1 \ --env-dir /path/to/workspace --清理up-first

This single command:

创建s VPC + subnet + IGW + 路由 table 创建s security group (NO inbound ports — SSM only) 创建s IAM 角色 with minimal 权限s (SSM + Parameter Store + Bedrock) Stores secrets in SSM Parameter Store (fetched at each 服务 启动 — rewritten on each 启动, never stored in repo or static images) Launches t4g.medium ARM64 instance with user-data bootstrap User-data 安装s Node.js 22 + OpenClaw + 配置s everything 运行s smoke test via SSM Saves all resource IDs to 部署-输出.json After 部署 Message the Telegram 机器人 — you'll 获取 a pAIring code 应用rove pAIring via SSM: aws ssm 启动-会话 --tar获取 --region us-east-1 sudo -u OpenClaw bash 导出 HOME=/home/OpenClaw OpenClaw pAIring 应用rove telegram

机器人 is live! ✅ Teardown # Using saved 输出: ./scripts/teardown.sh --from-输出 ./部署-输出.json --env-dir /path/to/workspace --yes

# Or by name (discovers via tags): ./scripts/teardown.sh --name starfish --region us-east-1 --env-dir /path/to/workspace --yes

模型 Support --模型 flag

Pass any 模型 string — it goes directly into OpenClaw.json as 模型.primary:

# Default (MiniMax M2.1 on Bedrock — no API key needed, uses IAM 角色) ./scripts/部署_minimal.sh --name starfish --region us-east-1

# Gemini Flash (needs GEMINI_API_KEY in .env.starfish) ./scripts/部署_minimal.sh --name starfish --region us-east-1 \ --模型 google/gemini-2.0-flash

AWS Bedrock

Bedrock IAM 权限s (bedrock:Invoke模型, bedrock:Invoke模型With响应流) are always 添加ed to the instance 角色 — regardless of which 模型 you choose. This means any 部署ed instance can use Bedrock 模型s out of the box via IAM 角色 凭证s (no API key needed).

Known Bedrock 模型 IDs:

模型 flag Description amazon-bedrock/minimax.minimax-m2.1 MiniMax M2.1 amazon-bedrock/minimax.minimax-m2 MiniMax M2 amazon-bedrock/deepseek.deepseek-r1 DeepSeek R1 amazon-bedrock/moonshotAI.kimi-k2.5 Kimi K2.5

Note: Bedrock 模型s must be enabled in your AWS account via the Bedrock console before use.

Gemini

If GEMINI_API_KEY is present in .env.starfish, it's stored in SSM and written to auth-性能分析s.json. If absent, it's simply skipped — no error.

.env.starfish TELEGRAM_机器人_令牌=... # Required — from @机器人Father GEMINI_API_KEY=... # Optional — from AIstudio.google.com (needed for Gemini 模型s)

Architecture (Minimal) ┌─────────────────────────────────────────────────────┐ │ VPC (10.50.0.0/16) │ │ ┌───────────────────────────────────────────────┐ │ │ │ Public Subnet (10.50.0.0/24) │ │ │ │ ┌─────────────────────────────────────────┐ │ │ │ │ │ EC2 t4g.medium (ARM64, 4GB) │ │ │ │ │ │ ┌───────────────────────────────────┐ │ │ │ │ │ │ │ OpenClaw Gateway │ │ │ │ │ │ │ │ • Node.js 22.14.0 │ │ │ │ │ │ │ │ • Any 模型 (Bedrock/Gemini/etc) │ │ │ │ │ │ │ │ • Telegram channel │ │ │ │ │ │ │ │ • 加密ed EBS (gp3, 20GB) │ │ │ │ │ │ │ └───────────────────────────────────┘ │ │ │ │ │ └─────────────────────────────────────────┘ │ │ │ └───────────────────────────────────────────────┘ │ └─────────────────────────────────────────────────────┘ ↑ ↓ SSM (no SSH/inbound) Outbound HTTPS only

Critical Lessons Learned (22 Issues)

These are baked into the 部署 script. See references/TROUBLESHOOTING.md for full detAIls.

Instance Sizing t4g.medium (4GB) required — t4g.small (2GB) OOMs during npm 安装 + gateway 启动up ARM64 — better price/performance than x86 Node.js Node 22+ required — OpenClaw 2026.x requires Node ≥22.12.0 Official tarball 安装 — NodeSource 设置up_22.x unreliable on AL2023 ARM64 git required — OpenClaw npm 安装 has git-based dependencies npm Use OpenClaw@latest — bare OpenClaw may resolve to placeholder package (0.0.1) Gat