MCP Server

Spider runs a built-in MCP server so AI coding assistants can query your crawl data directly. They read audit results, issues, and site structure without you copying anything out by hand.

What it exposes

The server provides 60+ project-scoped tools covering:

• Crawl lifecycle — start, stop, pause, resume, and wait on crawls
• Audit data — page reports, screenshots, site analysis, the site graph, sitemap, and robots.txt
• Issues — lists by page, severity, and type; broken links, redirects, duplicates, and quick wins
• Google data — imported Search Console top pages, queries, and metrics
• Insights — project health and trends, audit deltas, high-traffic pages at risk, link insights, and content opportunities

It also ships 8 SEO skills. These are guided, multi-step workflows an assistant can run end to end, rather than calling individual tools one at a time and stitching the results together itself:

• seo-audit — a full graded health check of the project
• seo-quick-wins — the highest-impact, lowest-effort fixes to ship first
• seo-striking-distance — turn near-page-one queries into a rank-gain plan
• seo-competitor — compare a page head-to-head against a competitor's
• seo-content-strategy — content gaps, topic clusters, and cannibalization
• seo-create-content — draft content grounded in the crawl and Knowledge
• seo-internal-linking — surface orphans and weak anchors, plan the links
• seo-progress-report — compare two crawls to prove fixes worked

Each skill drives the tools above in sequence, so the assistant runs the whole workflow from one instruction.

Connecting a client

Every server is bound to a single project with a mandatory --project argument. It won't start without one. Spider generates a ready-to-paste configuration for each client:

• Claude Code — add it with claude mcp add
• Cursor — add it to ~/.cursor/mcp.json
• Windsurf — add it to the Windsurf MCP config

Open a project's Connect AI agent action (or Settings → MCP Server) to copy the exact config.