Serpent is a desktop site crawler for technical SEO audits. It runs entirely on your machine, stores every crawl in a local SQLite file, sends no telemetry, and has no URL cap or licence server. It is MIT licensed and the source is public.
The part that matters for agent workflows is the built-in MCP server. While the app is open it serves Model Context Protocol on http://127.0.0.1:7777/mcp, bound to localhost and never exposed to the network. Point Claude Desktop or any MCP client at it and the assistant can start a crawl, poll its progress, list past crawls, read page-level SEO data, pull issues grouped by severity, and export the whole thing as CSV. No manual export step in the middle.
Crawling runs on a local engine by default. Sites behind bot protection can be routed through an optional Bright Data Web Unlocker zone with your own key, and JavaScript pages render through Electron’s bundled Chromium, so there is no separate headless browser to install. AI analysis is bring-your-own-key across OpenAI, Anthropic, Gemini, OpenRouter and local Ollama, and keys live in the OS keychain rather than a config file.
Installers for Windows, macOS and Linux are on the releases page. The SignPath Foundation signing application is still pending, so Windows will warn on first run until it clears.