SEO Score API — Model Context Protocol Server
Give Claude Desktop, Claude Code, Cursor, Windsurf, or any MCP-aware AI tool direct access to the SEO Score API: full audits, backlink data, monitoring, and history.
Quick start
npx -y seoscoreapi-mcp
Set your API key once and forget it:
export SEO_SCORE_API_KEY="ssa_your_key_here"
(Get a free key at https://seoscoreapi.com — no card required.)
Tools exposed
Not exposed, on purpose
- Report files (
GET /audit/export): the PDF is binary and an MCP tool result is text. Theaudittool returns the same data as JSON; use the REST API or the Python / Node SDK (audit_export/auditExport) for the file. - Turning auto-reload on. The API has no such endpoint for API keys: it authorizes
charges, so it is set from the dashboard only.
citation_auto_reloadreads it. - Account actions (
/signup,/verify,/upgrade,/rotate-key) and the public scoreboard opt-out.
Configuration
Claude Desktop
Edit ~/Library/Application Support/Claude/claude_desktop_config.json (Mac) or %APPDATA%\Claude\claude_desktop_config.json (Windows):
{
"mcpServers": {
"seoscoreapi": {
"command": "npx",
"args": ["-y", "seoscoreapi-mcp"],
"env": {
"SEO_SCORE_API_KEY": "ssa_your_key_here"
}
}
}
}
Restart Claude Desktop. You'll see the SEO Score tools in the tool picker.
Claude Code
claude mcp add seoscoreapi -e SEO_SCORE_API_KEY=ssa_your_key_here -- npx -y seoscoreapi-mcp
Cursor
Open Cursor Settings → MCP Servers → Add a new server with the same JSON shape as Claude Desktop above.
Windsurf
Settings → Cascade → MCP Servers → paste the same configuration.
Any other MCP client
Run npx -y seoscoreapi-mcp as a stdio transport. The server speaks standard MCP and exposes all tools in the table above.
Example prompts
Once installed, talk to your AI assistant naturally:
"Audit https://stripe.com and tell me the top 3 things to fix."
"Run a batch audit on stripe.com, square.com, and adyen.com, then compare their AI readability scores."
"What backlinks have we seen pointing at my-startup.com?"
"Check the score history for our marketing site over the last 30 days."
"Run a deep audit on my-startup.com as a SaaS site and list the high-severity findings."
"Run an accessibility audit on our pricing page and list the critical WCAG failures."
"Crawl my-startup.com and show me the broken links and orphan pages."
"Build starter prompts for 'SEO audit API', create a citation tracker for Acme, and run it."
The tool the AI picks is shown in your client; you can always ask "what tools did you use?" to see the trace.
Why MCP?
MCP lets the AI reach data it wouldn't have — your actual site SEO scores, the backlink graph, your monitoring history — without any custom plumbing. Every prompt that says "audit this URL" goes from idea to result in one round-trip. Same source of truth as the REST API; no copying scores between tools.
Honest framing
The backlinks tool returns an audit-fed sample, not a comprehensive backlink index. Every response includes a data_caveat explaining exactly what the dataset is. We are not Ahrefs and don't pretend to be — we expose the unique, growing graph our customers' audits build for free.
Source
The MCP server is open source: https://github.com/SeoScoreAPI/seoscoreapi-mcp
License
MIT