Raptor or Screaming Frog?
Screaming Frog’s SEO Spider is the established desktop crawler, 24 major versions in, and it does a great deal Raptor doesn’t. Raptor is for a different part of the job: asking about a crawl in Claude, failing a build in CI, and handing a client a report they can read.
Everything about Screaming Frog here was checked against its own site, pricing page and user guide on . The pages are listed at the end.
The short version
Screaming Frog is the better fit when you
- want to explore a crawl in a desktop interface, tab by tab and filter by filter
- need custom extraction with XPath, CSS Path or regex
- want Google Analytics and Search Console data beside each URL
- audit accessibility, spelling and grammar, or near-duplicate content
- crawl a list of URLs, such as a migration’s redirects
- need an installer today, for Windows, macOS or Linux
Raptor is the better fit when you
- want to ask about a crawl in Claude, with no licence for the MCP server
- want the answer first: findings ranked by what to fix, with a fix brief for each
- want a crawl to fail a CI build when the score drops or a finding appears
- need one HTML report to send a client, with your agency’s brand on it
- want the code under the MIT licence, with no licence key and no page cap
Side by side
Raptor first, Screaming Frog second, row by row. Where we looked for something and didn’t find it, the cell says so.
| What | Raptor | Screaming Frog |
|---|---|---|
| Cost and licence | ||
| Price | RaptorFree. The page budget is a flag, --max-pages, 500 by default. | Screaming FrogFree up to 500 URLs per crawl. A licence is £199, $279 or €245 a year per person, less from five licences.From SEO Spider pricing |
| Licence | RaptorMIT. The public repository isn’t open yet. | Screaming FrogA commercial end-user licence, one person per key. When it expires, the app goes back to the free version.From SEO Spider FAQ |
| How you work with it | ||
| Interface | RaptorA command line, an MCP server, a REST API and the HTML report. No desktop interface. | Screaming FrogA desktop app, plus a command line that can launch, crawl, save and export without the interface. There is no API.From User guide: general |
| Install | RaptorBuild from source with Rust, rustc 1.88 or newer. A release workflow is ready to build binaries for Linux, macOS and Windows once the repository opens. | Screaming FrogInstallers for Windows, macOS and Linux. A 64-bit system is required.From SEO Spider FAQ |
| AI assistants | RaptorAn MCP server with nine tools in every build, over stdio, for Claude Code, Claude Desktop or any MCP client. | Screaming FrogAn MCP server since version 24.0 (May 2026), with 29 tools over stdio or Streamable HTTP. It needs a paid licence and the default database storage mode.From User guide: MCP server |
| AI while crawling | RaptorNone. Claude does the reading, through the MCP server. | Screaming FrogUp to 100 custom prompts per crawl with OpenAI, Gemini, Ollama or Anthropic, written into the crawl’s columns. Paid.From User guide: configuration |
| Automation | RaptorExit codes, --min-score and --fail-on to fail a build, and JSON on stdout for scripts. | Screaming FrogScheduled crawls in the app, which export headless (paid), and the command line for anything else.From User guide: general |
| What it checks | ||
| Checks | Raptor128 checks in six areas, each with what it looks for, why it matters, how to fix it and what it can’t know. | Screaming Frog“Over 300 SEO issues, warnings and opportunities”, in its own words. The two count differently, so the numbers don’t compare.From SEO Spider overview |
| JavaScript | RaptorRaw HTML by default. Local Chrome through Playwright in a build with the playwright feature. | Screaming FrogRaw HTML by default. Headless Chromium rendering is a paid feature.From User guide: configuration |
| Integrations | RaptorGoogle PageSpeed for Core Web Vitals, when you ask for it. | Screaming FrogGoogle Analytics, Search Console, PageSpeed Insights, and link metrics from Majestic, Ahrefs and Moz. Paid.From SEO Spider overview |
| Custom extraction | RaptorNone. | Screaming FrogXPath, CSS Path and regex, and custom JavaScript run while crawling. Paid.From SEO Spider overview |
| Internal links | RaptorPageRank over the crawl’s links, with orphans, hubs and silos, from raptor link or Claude. | Screaming FrogA Link Score for each URL, and force-directed and tree diagrams of the site’s structure.From SEO Spider overview |
| Comparing crawls | RaptorScore change, new and fixed problems, and whether two crawls are like for like, through the MCP server. | Screaming FrogCrawl comparison, and staging against production with URL mapping. Paid.From SEO Spider overview |
| What you get, and where it lives | ||
| Report | RaptorOne self-contained HTML file: the grade and why, what to fix first, and every finding and page. It prints to PDF and can carry your brand. | Screaming FrogExports to CSV, Excel and Google Sheets, and a Data Studio report through Sheets (paid). We found no HTML or PDF audit report in its user guide.From User guide: general |
| Your data | RaptorOn your machine. No account and no telemetry. | Screaming FrogCrawl data stays on your machine, in a local database by default. Its privacy policy covers the licence, crash and debug data it collects.From SEO Spider FAQ |
| Big sites | RaptorBounded by --max-pages and --depth. There’s no published guidance on the largest crawl. | Screaming Frog16GB of RAM and an SSD recommended to crawl millions of URLs.From SEO Spider FAQ |
In Claude, both ways
Both have MCP servers now, so both can be asked about a crawl. They hand Claude different things: Screaming Frog’s tools give it the crawl’s data and exports to work through, Raptor’s give it a verdict and the findings behind it.
Raptor’s nine tools
- raptor_crawl
- raptor_status
- raptor_cancel
- raptor_summary
- raptor_findings
- raptor_query
- raptor_inspect
- raptor_compare
- raptor_link_graph
In every build, with no licence. The summary and the ranked findings are the verdict; query, inspect, compare and the link graph are the evidence. What each one returns
Screaming Frog’s, some of them
- sf_crawl
- sf_crawl_progress
- sf_load_crawl
- sf_generate_report
- sf_generate_bulk_export
- sf_url_info
- sf_url_links
- sf_run_node_js_script
Eight of its 29 tools. The rest list and load crawls, export data, read and write files, and install npm packages; the Node.js and file tools are off until you turn them on.
From User guide: MCP serverWhere Raptor is thinner
Version 0.3.0 against 24 major versions. These are the gaps worth knowing before you choose.
No desktop interface
You explore a crawl in the report, the JSON or Claude. Screaming Frog’s interface is built for filtering a crawl tab by tab.
No list mode
Raptor crawls outward from one start address, or audits a single page. Screaming Frog can crawl a list of URLs, which is how a migration’s redirects get checked.
No accessibility, spelling or near-duplicate checks
Screaming Frog audits accessibility with the open-source axe rules, spelling and grammar in over 25 languages, and near-duplicate content. Raptor finds exact duplicates only.
Nothing from Google Analytics or Search Console
Raptor knows what the crawl sees and nothing else: no traffic, no impressions, no index status from Google.
Where the facts come from
Each row about Screaming Frog names one of these pages, read on 18 October 2026. Prices and features change, so if something here no longer matches, the page has gone out of date, and these links are where to check.