The read-only SEO connectors leave the actual fix to you: you get keyword tables and crawl reports and then go do the work. Rampify's bet is that the same connection should also write the title, the meta and the JSON-LD and open the PR, so marketing findings end up with the same paper trail as engineering work — a spec, affected URLs, and the commits that closed it.
A remote MCP server for SEO and AI-answer visibility work, connected over OAuth with no API keys to manage. It pulls keyword data through DataForSEO and your Google Search Console performance, adds a live crawl of your own site, and then acts on what it finds inside the coding agent you already use.
- A crawl and scan of your own site covering meta, schema, content and Core Web Vitals, so findings are grounded in the pages you actually ship
- Keyword volume and trend lookups plus your Search Console performance, without bolting on separate data servers
- Generated titles, meta tags and JSON-LD schema written against that data rather than from a template
- Discovery research that runs as your buyer and shows where you are invisible when people ask AI assistants about your category
- Findings turned into feature specs with the affected URLs and tasks attached, linkable to the commits that resolve them — and a PR opened against your repo, or a publish through your CMS's own MCP
A Rampify account, authorized over OAuth on first connect. Nothing to install and nothing to keep updated: the server is remote at `https://www.rampify.dev/api/mcp` over Streamable HTTP, and new tools appear on their own. Cursor and VS Code offer a one-click add; for Claude Code it is one command, `claude mcp add --transport http rampify https://www.rampify.dev/api/mcp`; claude.ai takes it as a custom connector by URL. The older `@rampify/mcp-server` npm package is deprecated in favour of the hosted server, and the source in the repository is kept for reference only.
One command — claude mcp add --transport http rampify https://www.rampify.dev/api/mcp