DEV Community

Benneth Uzochukwu
Benneth Uzochukwu

Posted on

I Built an MCP Server That Gives AI Assistants Direct Access to 16 SEO & Marketing Tools

#mcp #ai #seo #typescript

Every SEO workflow I had involved the same dance: open the dashboard, click around, copy data, paste it into my AI assistant, ask a question, get an answer, go back to the dashboard. Repeat 50 times a day.

So I built an MCP server that eliminates the middleman entirely.

What is MCP?

Model Context Protocol (MCP) is an open standard that lets AI assistants (Claude, Cursor, VS Code Copilot, etc.) call external tools directly. Instead of copy-pasting data into a chat, the AI can reach out and grab what it needs.

Think of it like giving your AI assistant hands.

What SearchAtlas MCP Server Does

It connects any MCP-compatible AI client to the SearchAtlas platform — 10 specialized AI agents and 6 management tools, all accessible through natural conversation.

Instead of this:

Opens dashboard → navigates to keyword tool → types query → copies results → pastes into Claude → asks for analysis

You just say:

"Find long-tail keywords for project management software with low difficulty"

And the AI calls the right tool, gets the data, and analyzes it — all in one step.

The 16 Tools

🤖 10 AI Agents

Tool What It Does
searchatlas_orchestrator Routes your query to the best specialist
searchatlas_otto_seo Technical SEO fixes, schema markup, optimizations
searchatlas_ppc Google Ads campaigns, bids, performance analysis
searchatlas_content Blog posts, landing pages, optimized copy
searchatlas_site_explorer Site audits, backlinks, competitive intelligence
searchatlas_gbp Google Business Profile, reviews, local SEO
searchatlas_authority_building Link building, digital PR, outreach
searchatlas_llm_visibility Track how AI models reference your brand
searchatlas_keywords Search volume, difficulty, SERP analysis
searchatlas_website_studio Page builder, layouts, site structure

🛠️ 6 Management Tools

Tool What It Does
searchatlas_list_projects List your projects
searchatlas_create_project Create a project by domain
searchatlas_list_conversations Browse past chat sessions
searchatlas_list_artifacts Find generated content and reports
searchatlas_list_playbooks Browse automation recipes
searchatlas_run_playbook Execute a playbook on any project

Setup Takes 60 Seconds

1. Install & Login 

npm install -g searchatlas-mcp-server
searchatlas login
Enter fullscreen mode Exit fullscreen mode

The login command opens your browser, validates your token, saves it, and auto-configures your MCP clients. Done.

2. Add to Your AI Client

Claude Code:

claude mcp add searchatlas -e SEARCHATLAS_TOKEN=your-token -- npx -y searchatlas-mcp-server
Enter fullscreen mode Exit fullscreen mode

Cursor (.cursor/mcp.json):

{
  "mcpServers": {
    "searchatlas": {
      "command": "npx",
      "args": ["-y", "searchatlas-mcp-server"],
      "env": {
        "SEARCHATLAS_TOKEN": "your-token"
      }
    }
  }
}
Enter fullscreen mode Exit fullscreen mode

Works with Claude Code, Cursor, Claude Desktop, VS Code, Windsurf, and Zed.

3. Verify 

searchatlas check
Enter fullscreen mode Exit fullscreen mode 
✓ Credential source: ~/.searchatlasrc
✓ Config loaded successfully
✓ JWT structure valid (expires in 12 days) — user 42
✓ API reachable and authenticated

All checks passed — you're ready to go!
Enter fullscreen mode Exit fullscreen mode

Real Usage Examples

Once connected, just talk naturally:

"Run a technical SEO audit on example.com"
"Write a blog post about headless CMS best practices"
"What are the top backlink opportunities for my site?"
"Show my projects and run the content optimization playbook on the first one"
"How is my brand showing up in AI model responses?"
Enter fullscreen mode Exit fullscreen mode

The orchestrator agent automatically routes your request to the right specialist. You don't need to think about which tool to use.

How It Works Under the Hood

The server uses stdio transport — your MCP client spawns it as a local process and communicates via JSON-RPC over stdin/stdout. Agent tools use Server-Sent Events (SSE) streaming to collect responses from SearchAtlas's backend.

No HTTP server to deploy. No ports to open. No Docker required (though a Dockerfile is included if you want it).

The architecture:

Your AI Client (Claude / Cursor / VS Code)
        ↕  JSON-RPC over stdio
SearchAtlas MCP Server (local Node.js process)
        ↕  HTTPS + SSE
SearchAtlas API (10 specialized AI agents)
Enter fullscreen mode Exit fullscreen mode

Built With

MIT licensed. PRs welcome.

Links

If you work in SEO or digital marketing and use AI assistants daily, give it a try. The 60-second setup is worth it.

Top comments (0)

Some comments may only be visible to logged-in visitors. Sign in to view all comments.