The Model Context Protocol is an open standard that lets AI assistants securely connect to external tools and data. An MCP server exposes capabilities; clients like Claude Desktop, Claude Code and Cursor can call them mid-conversation.

Which clients work with the EvidentlyAEO MCP server?

Any MCP-compatible client. Claude Desktop, Claude Code and Cursor are the most common; the server speaks standard MCP over HTTP at evidentlyaeo.com/mcp.

What capabilities are exposed?

24 analytical tools across 11 domains: dashboard KPIs, queries, topics, citations, brand reputation and sentiment, recommendations, domain readiness, content scoring, ad intelligence, shopping intelligence, and brands. All are read-only.

Access is authenticated with a bearer API key scoped to your workspace, over HTTPS. The server only exposes read access to your own tracked data.

Do I need a specific plan?

MCP access is included with plans that have API access. Check the pricing page or talk to us for details on your tier.

Ad Intelligence: see how your brand shows up in ChatGPT ads. Learn more →

EvidentlyAEO

MCP Server

Your AEO data, inside your AI client.

The full EvidentlyAEO dataset exposed as 24 Model Context Protocol tools. Ask Claude or Cursor about your visibility, citations and competitors, and get answers from live data.

Start Free Trial

Add the serverstreamable HTTP

claude_desktop_config.json

{
  "mcpServers": {
    "evidently-aeo": {
      "url": "https://evidentlyaeo.com/mcp",
      "headers": {
        "Authorization": "Bearer <YOUR_API_KEY>"
      }
    }
  }
}

Claude DesktopClaude CodeCursorAny MCP client

01Connect

One config block, no SDK.

The server speaks standard Model Context Protocol over streamable HTTP. Paste a URL and a bearer token into your client's config, and your AEO data is in the conversation. Nothing to install, no integration project.

Works in Claude Desktop, Claude Code, Cursor, or any MCP client
Authenticates with a workspace API key you generate yourself
Composes with your other MCP servers in the same conversation

In your MCP clientexample

Which sources cite my competitors but not me?

citations_competitor_gap done

{ "brandId": "…", "startDate": "2026-07-22" }

Three domains cite competitors on your tracked topics but have never cited you: g2.com, runrepeat.com, and reddit.com. The largest gap is on comparison queries.

02Ask

The model picks the tool.

Ask in plain English. The client reads the tool descriptions, calls the right one with the right arguments, and answers from live data. No dashboard navigation, no CSV export, no copy-paste into a prompt.

Every answer traces back to a named tool call you can inspect
Follow-up questions chain tools without starting over
Cross-reference your AEO data against anything else in context

Tool catalog24 tools

dashboard

dashboard_get_summarydashboard_list_competitorsdashboard_llm_breakdowndashboard_get_action_items

queries

queries_summaryqueries_trendqueries_competitor_overlapqueries_collector_breakdown

citations

citations_top_sourcescitations_source_detailcitations_competitor_gapcitations_trend

brand_reputation

brand_reputation_summarybrand_reputation_attributesbrand_reputation_sentiment_trend

recommendations

recommendations_listrecommendations_get_detail

shopping_intelligence

shopping_intelligence_analyzeshopping_intelligence_product_detail

topics

topics_performance

+ domain_readiness, content_scorer, ad_intelligence, brands

03Coverage

24 typed tools across 11 domains.

Every domain the platform measures is exposed as its own documented, typed tool. The model selects precisely, and you get structured data back instead of scraped prose.

Visibility, citations, queries, topics and competitor breakdowns
Brand reputation, sentiment trends, and attribute-level scoring
Recommendations, domain readiness audits, and content scoring
Ad intelligence and shopping intelligence assessments

Key scopesRead-only

read:dashboardread:queriesread:citationsread:recommendationsread:brandsread:domainread:contentread:productsread:ads

metric-dictionary

Canonical definitions for Visibility, Share of Answer, Presence Rate, sentiment scoring, blind vs. brand queries, and null-value rules.

prompts

aeo-expertquick-summarydata-only

04Grounded

Scoped keys and defined metrics.

Every key is scoped and read-only, and every request validates brand ownership before returning a row. A metric-dictionary resource ships alongside the tools so models interpret Share of Answer and sentiment the way your team defines them.