Screaming Frog MCP (headless)
Runs Screaming Frog crawls with the app closed, through nine read-and-export tools and nothing else.
Add it to your client
Communityuv tool install screaming-frog-mcp
{
"mcpServers": {
"screaming-frog-mcp": {
"command": "uv",
"args": [
"tool",
"install",
"screaming-frog-mcp"
]
}
}
}- Transport
- stdio
- Runtime
- Python
- Auth
- Not required
- Tools
- 9
- License
- MIT
- Stars
- 79
Screaming Frog ships its own MCP server in version 24, and that one does more: screenshots, crawl comparison, an optional Node runtime. This is the deliberately smaller alternative, built for crawls nobody is watching. There is no script runner, no package installation and no filesystem write access, and an allowlist environment variable can restrict which domains an agent is permitted to crawl at all. The tool that earns its place is aggregate_crawl_data, which computes counts and group-by breakdowns inside the server, so answering how many pages return a 404 costs one number rather than a full CSV export read into context. It drives the Screaming Frog command line and its saved crawl database directly, so a licensed install is required.
Similar MCP servers
Firecrawl MCP
MCP serverTurns web pages and whole sites into markdown an agent can read.
Hyperbrowser MCP Server
MCP serverCloud browsers for agents: scrape, crawl, extract and run browser agents without local Chrome.
Graphlit MCP Server
MCP serverIngests Slack, Notion, Drive, email, Jira and web pages into a Graphlit project, then searches it.

Comments(0)
Sign in to comment