A bridge to the Internet Archive's Wayback Machine that works without authentication (though you can add credentials for higher SPN2 rate limits). You get seven tools: save URLs, retrieve archived snapshots, search the CDX API for historical captures, check availability via sparkline, list screenshots, compare snapshots, and clear cache. Built with a 15 req/min token bucket rate limiter, automatic Retry-After handling for 429s, and tiered cache TTLs (24h for immutable snapshot content, 1h for CDX/availability, 30min for saves). Reach for this when you need to programmatically archive pages, pull historical versions of a site, or diff snapshots across time. Ships as both an MCP server and a CLI, written in TypeScript with strict Zod validation and full test coverage.
claude mcp add --transport stdio mearman-mcp-wayback-machine uvx mcp-wayback-machine