MCP SERVER → rutt.io

Rutt for AI coding assistants

The official Model Context Protocol server. Give Claude Code, Cursor, Windsurf, or any MCP-aware tool full read access to your AI visibility, bot traffic, rankings, and brand context.

Endpoint https://rutt.io/api/mcp

What is the Rutt MCP server?

Model Context Protocol is an open standard for connecting AI assistants to external systems. The Rutt MCP server is hosted at https://rutt.io/api/mcp — nothing to install. Point your assistant at the URL with your API key and you get 40 tools and 4 reference resources.

Once connected, your assistant can answer questions like:

“How is acme.com performing on ChatGPT vs Claude this month?”
“Which tracked keywords lost rank this week?”
“What prompts do our competitors win where we lose?”
“What does this brand sound like — write the blog post in that voice.”

Every tool response includes unit hints, a one-sentence interpretation, and suggested next tool calls — so the assistant never has to guess what the data means.

Connect

First, create an API key at /integrations/api-keys. Keep it scoped to read unless you specifically want the assistant to write data back.

Then point your assistant at:

https://rutt.io/api/mcp — Authorization: Bearer rutt_xxx

Claude Code

One command:

claude mcp add --transport http rutt https://rutt.io/api/mcp \
  --header "Authorization: Bearer rutt_your_api_key_here"

Cursor

Edit ~/.cursor/mcp.json (or the project-local .cursor/mcp.json):

{
  "mcpServers": {
    "rutt": {
      "url": "https://rutt.io/api/mcp",
      "headers": {
        "Authorization": "Bearer rutt_your_api_key_here"
      }
    }
  }
}

Windsurf

Edit ~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "rutt": {
      "serverUrl": "https://rutt.io/api/mcp",
      "headers": {
        "Authorization": "Bearer rutt_your_api_key_here"
      }
    }
  }
}

Claude.ai & ChatGPT

Claude.ai and ChatGPT's connector UIs currently expect OAuth, which Rutt doesn't implement yet. Until they do, pass your API key as a query parameter and skip the auth step. Open Settings → Connectors → Add custom connector and paste:

https://rutt.io/api/mcp?api_key=rutt_your_api_key_here

Leave OAuth Client ID and Client Secret blank.

Security note: An API key in a URL is more exposed than one in a header — it can appear in browser history, server logs, and Claude.ai's analytics. For this connector, create a dedicated read-only key, name it “Claude.ai connector” or similar at /integrations/api-keys, and revoke it the moment you stop using it. Don’t paste the URL in public chat logs or screenshots.

Test it: curl -X POST https://rutt.io/api/mcp -H "Authorization: Bearer rutt_xxx" -H "Content-Type: application/json" -d '{"jsonrpc":"2.0","id":1,"method":"tools/list"}'

Tools

44 tools across 10 product areas. All read-only except where marked write.

Sites

Tool	Description
`rutt_list_sites`	List all sites in the team.
`rutt_get_site`	Full site detail including brand profile and crawl config.
`rutt_get_site_summary`	One-call dashboard snapshot: visibility, bots, rankings, open actions. Best first call.

Brand & Voice

Tool	Description
`rutt_get_brand_profile`	Industry, tone, target audience, key products, guidelines. Call before generating content.
`rutt_list_site_pages`	Crawled pages with SEO scores and context flags.
`rutt_get_page_seo_report`	Detailed SEO audit for one page.

AI Visibility

Tool	Description
`rutt_get_visibility_overview`	Per-model scores, mention rates, top competitors, top cited sources.
`rutt_get_visibility_trend`	Daily visibility score series; optional model filter.
`rutt_list_monitoring_prompts`	Prompts being checked across AI models.
`rutt_get_monitoring_prompt`	Single prompt with recent mention/recommendation rates.
`rutt_get_prompt_results`	Individual check results for one prompt.
`rutt_get_prompt_result_detail`	Full AI response text for one check.
`rutt_add_monitoring_prompt` write	Create a new monitoring prompt.

Bot Analytics

Tool	Description
`rutt_get_bot_traffic`	Human vs AI vs search vs other bot totals + AI share %.
`rutt_get_traffic_by_crawler`	Per-crawler breakdown with intent (training / user_action / search).
`rutt_get_top_crawled_pages`	URLs AI crawlers hit most. Filter by bot.
`rutt_get_ai_referrers`	Hosts sending traffic — spots ChatGPT/Perplexity referrals.
`rutt_get_traffic_breakdown`	Pageview breakdown by visitor type.

Rank Tracker

Tool	Description
`rutt_list_tracked_keywords`	Keywords with latest SERP position + AI Overview flag.
`rutt_get_keyword`	One keyword with latest rank data.
`rutt_get_keyword_rank_history`	Daily rank history for a keyword.
`rutt_get_rank_movers`	Biggest gainers and losers over the window.
`rutt_get_ai_overview_keywords`	Keywords where Google AI Overview triggered + cited sources.
`rutt_add_tracked_keyword` write	Track a new keyword.

Keyword Research

Tool	Description
`rutt_get_keyword_metrics`	Bulk volume/KD/CPC/intent for any keyword list (up to 700). Doesn't require tracking.
`rutt_research_keywords` write	Run keyword research from a seed term. Persists to the team — appears in /keywords. Async.
`rutt_get_keyword_research`	Poll for the completed result of a research run.
`rutt_list_keyword_research`	Past keyword research runs for a site.

Competitors / SEO Rivals

Tool	Description
`rutt_list_competitors`	Competitors with DR, backlinks, organic traffic.
`rutt_get_competitor`	Single competitor with SEO metrics.
`rutt_get_competitor_pages`	Discovered pages for a competitor.
`rutt_analyze_domain` write	Analyze any domain — DR, backlinks, traffic. Auto-adds it as a tracked competitor. Async.
`rutt_get_visibility_gap`	Prompts where competitors win and we don't. High-signal for content strategy.
`rutt_list_fanout_queries`	Sub-questions LLMs derive from monitoring prompts.

Weekly Plan

Tool	Description
`rutt_get_current_weekly_plan`	This week's AI-generated day-by-day action plan.
`rutt_list_weekly_plans`	Past weekly plans.
`rutt_get_weekly_plan_items`	Items for a specific plan.
`rutt_mark_action_complete` write	Mark a plan item complete.

Reports

Tool	Description
`rutt_list_reports`	Generated reports for a site.
`rutt_get_report`	Report metadata + public share URL.

Reference / discovery

Tool	Description
`rutt_get_reference_models`	AI models Rutt monitors.
`rutt_get_reference_countries`	Allowed monitoring countries (plan-gated).
`rutt_get_reference_crawlers`	Known crawler signatures + intent.
`rutt_get_glossary`	Definitions for Rutt-specific terms. Call once at session start.
`rutt_get_plan_limits`	Current team's plan and feature limits.

Resources

In addition to tools, the server exposes resources — URI-addressable reference data your assistant can auto-load at session start:

URI	Contents
`rutt://glossary`	Rutt-specific terminology so the assistant interprets data correctly.
`rutt://models`	AI models monitored (chatgpt, claude, gemini, perplexity).
`rutt://crawlers`	Known bot signatures with company + intent (training / user_action / search).
`rutt://plan-limits`	Current team's plan and feature limits.

The AI envelope

Every tool result is wrapped in an envelope that makes the data self-describing. The same envelope is available on the plain REST API by adding ?ai=1 or the header Accept: application/vnd.rutt.ai+json.

{
  "data": { ... },
  "meta": {
    "window": { "start": "2026-04-12", "end": "2026-05-12", "label": "last 30 days" },
    "site": { "id": 12, "domain": "acme.com" },
    "unit_hints": {
      "score": "0-100, % of monitored prompts where this site appeared in AI answers"
    }
  },
  "interpretation": "Strongest on ChatGPT (71/100), weakest on Claude (58/100). Top competitor mentioned: Notion.",
  "next_actions": [
    {
      "tool": "rutt_get_visibility_gap",
      "args": { "site_id": 12 },
      "why": "Find prompts where competitors win"
    }
  ]
}

unit_hints describe what each metric means and its range, so the LLM never has to guess. interpretation is a one-sentence natural-language summary generated server-side from the same data. next_actions suggest follow-up tool calls so the assistant can chain without re-asking the user.