spotinfo

概述

spotinfo 是一个全面的 CLI 工具和 MCP 服务器，提供有关 AWS EC2 Spot 实例的详细信息。它提供实时放置分数（1-10 分）、定价数据和中断频率分析。该工具具有高级过滤功能，包括基于正则表达式的实例类型匹配，以及按 vCPU、内存、价格、地区和放置分数进行多维度过滤。它支持多种输出格式，包括可视化表格、JSON 和 CSV，并在 AWS API 不可用时提供优雅的后备方案。MCP 集成允许 AI 助手通过自然语言直接查询 AWS Spot 实例数据。

试试问 AI

装完之后，这里有 5 个你可以让 AI 做的事：

• query_spot_instances

  Query AWS EC2 Spot instances with filtering by type, region, price, and placement scores

  试试:Find cheapest t3 instances with placement score >7试试:Compare m5.large prices across US regions试试:Show high-reliability instances with 4 vCPU and 16GB memory under $0.30
• get_placement_scores

  Get real-time placement scores (1-10 scale) for spot instances

  试试:What's the placement score for t3.micro instances in us-east-1试试:Compare placement scores for m5.* instances across availability zones
• compare_spot_prices

  Compare spot instance prices across different regions

  试试:Compare c5.xlarge prices between eu-west-1 and us-west-2试试:Show the price difference for r6g instances between all regions
• filter_instances

  Filter spot instances by vCPU, memory, price, and other attributes

  试试:Find instances with at least 8 vCPU and 32GB memory试试:Show instances with price under $0.20 in us-east-1试试:Filter by instance type pattern c5.* with minimum score 8
• export_spot_data

  Export spot instance data in JSON or CSV format for automation

  试试:Export all m5 instances with scores to JSON试试:Get spot price data in CSV format for c5.* instances

说明：Tool names inferred from MCP integration section and usage examples since explicit tool names aren't directly documented

可对比工具

安装

# macOS 使用 Homebrew
brew tap alexei-led/tap
brew install alexei-led/tap/spotinfo

# Linux/Windows：从发布版下载
curl -L https://github.com/alexei-led/spotinfo/releases/latest/download/spotinfo_linux_amd64.tar.gz | tar xz

# Docker
docker pull ghcr.io/alexei-led/spotinfo:latest

{
  "mcpServers": {
    "spotinfo": {
      "command": "spotinfo",
      "args": ["--mcp"]
    }
  }
}

FAQ

使用 spotinfo 需要 AWS 凭证吗？

不需要，AWS 凭证是可选的。没有凭证，您仍然可以访问静态的 Spot 顾问数据和定价，但实时价格回退和放置分数将不可用。

放置分数是什么意思？

放置分数（1-10 分）代表您的 Spot 实例请求的成功概率，考虑您整个请求的上下文。分数越高表示成功机会越大，当您允许更多实例类型灵活性时，分数会提高。

spotinfo 对比