Screaming Frog SEO Spider MCP Server — Claude Code MCP server: setup
Run Screaming Frog crawls and query saved crawl data through MCP without keeping the GUI open. Useful for exports and audits that run unattended.
claude mcp add --env SF_CLI_PATH=<SF_CLI_PATH> --transport stdio screaming-frog-seo-spider-mcp-server-headless -- uvx screaming-frog-mcpWhat Screaming Frog SEO Spider MCP Server (headless) does
This MCP server controls Screaming Frog SEO Spider through its command-line interface. It can start crawls, export saved crawl data, filter exported rows, and calculate counts or distributions on the server. Its tools are limited to reading and exporting data; this is not a cloud crawler.
You need Screaming Frog SEO Spider with a valid license, Python 3.10+, and database storage mode. The desktop app locks the crawl database, so close it before asking the server to read saved data. Crawls kept in memory storage are outside the reach of the database-reading tools.
Screaming Frog starts only when a tool needs it, and the database is released between calls. That suits scheduled checks, though the first response for a crawl can take longer than a follow-up in a session where the app stays open. This is a community project, separate from Screaming Frog.
What it helps with
- start a site crawl from the command line
- export data from a saved crawl
- filter export rows with a regular expression
- count error distributions in crawl data
- check the license and database lock before a crawl
Example prompts for the agent
- Run a site crawl with
crawl_siteand inspect the result. - List saved crawls, then export the latest one.
- Find rows for 404 errors in the crawl export.
How to install Screaming Frog SEO Spider MCP Server (headless)
Platform CLI
claude mcp add --env SF_CLI_PATH=<SF_CLI_PATH> --transport stdio screaming-frog-seo-spider-mcp-server-headless -- uvx screaming-frog-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from screaming-frog-mcp <command>
Platform CLI
codex mcp add screaming-frog-seo-spider-mcp-server-headless --env SF_CLI_PATH=<SF_CLI_PATH> -- uvx screaming-frog-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from screaming-frog-mcp <command>
Config file
{
"mcpServers": {
"screaming-frog-seo-spider-mcp-server-headless": {
"command": "uvx",
"args": [
"screaming-frog-mcp"
],
"env": {
"SF_CLI_PATH": "<SF_CLI_PATH>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from screaming-frog-mcp <command>
Config file
{
"servers": {
"screaming-frog-seo-spider-mcp-server-headless": {
"type": "stdio",
"command": "uvx",
"args": [
"screaming-frog-mcp"
],
"env": {
"SF_CLI_PATH": "<SF_CLI_PATH>"
}
}
}
}Replace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from screaming-frog-mcp <command>
Platform CLI
gemini mcp add -e SF_CLI_PATH=<SF_CLI_PATH> screaming-frog-seo-spider-mcp-server-headless uvx screaming-frog-mcpReplace <…> with your own values. uvx runs the command named after the package; if the README names another one, use uvx --from screaming-frog-mcp <command>
Step by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Screaming Frog SEO Spider MCP Server (headless)
How do I install Screaming Frog SEO Spider MCP Server in Claude Code?
Add it with claude mcp add screaming-frog-mcp -- uvx screaming-frog-mcp. To set the CLI path explicitly, use claude mcp add --env SF_CLI_PATH=<SF_CLI_PATH> --transport stdio screaming-frog-seo-spider-mcp-server-headless -- uvx screaming-frog-mcp.
What does Screaming Frog SEO Spider MCP Server do?
It runs Screaming Frog SEO Spider crawls and exposes saved crawl data for export and filtering through MCP. The Screaming Frog desktop app must be closed while the server accesses the database.
Does Screaming Frog need to be closed before querying a crawl?
Yes. Its GUI locks the database, so close the app before the MCP server reads saved crawl data. Database storage mode is required for those data tools.
