generated: '2026-08-13'
method: generated
source: >-
  https://www.screamingfrog.co.uk/guides/mcp-server/ ,
  https://www.screamingfrog.co.uk/seo-spider/user-guide/general/#command-line
provider: Screaming Frog
providerId: screaming-frog
description: >-
  Packaged Agent Skills for the two machine-callable Screaming Frog surfaces. Screaming
  Frog publishes no OpenAPI, so these are not grounded in operationIds — they are grounded
  in the provider's own published MCP tool names and CLI flags, each transcribed verbatim
  from the pages cited above. No tool name, parameter or flag in these skills was invented.
grounding: mcp-tools-and-cli-flags
provider_published_skills: none
provider_published_skills_note: >-
  No AGENTS.md, skills/ directory or agent instruction file is published by Screaming Frog.
  Searched the docs host and the GitHub org; nothing found.

skills:
  - name: screaming-frog-technical-seo-audit
    file: screaming-frog-technical-seo-audit.md
    surface: mcp
    description: Crawl a site and produce a prioritised issue report without overflowing the context window.
  - name: screaming-frog-broken-link-report
    file: screaming-frog-broken-link-report.md
    surface: mcp
    description: Export broken links together with their inlinks, so the output is a fix list.
  - name: screaming-frog-scheduled-headless-crawl
    file: screaming-frog-scheduled-headless-crawl.md
    surface: cli
    description: Unattended headless crawl for cron/CI, including crawl-over-crawl comparison.

prerequisites:
  - A licensed Screaming Frog SEO Spider v24+ installed on the machine running the agent.
  - Database storage mode (the default) for MCP and for crawl comparison.
  - For MCP, the Node.js runtime accepted under 'File > Settings > MCP Server' if script tools are used.

maintainers:
  - FN: Kin Lane
    email: kin@apievangelist.com