trakkr-mcp 0.3.0

MCP server for Trakkr — AI visibility monitoring for ChatGPT, Perplexity, Gemini, and more

trakkr-mcp

MCP server for Trakkr — AI visibility monitoring.

Connect your AI assistant (Claude Code, Claude Desktop, Cursor, VS Code, Codex CLI, Windsurf, Zed, Cline, Continue, and any other MCP client) to Trakkr and query your brand's AI search visibility data conversationally.

Quick Start

The fastest path is uv — uvx auto-installs and runs the server with zero setup:

# One-time, if you don't have uv:
brew install uv

# Set your API key (get one at trakkr.ai/exports/api-keys):
export TRAKKR_API_KEY="sk_live_your_key_here"

# Optional, run the server directly to verify it works:
uvx trakkr-mcp

You usually don't run uvx trakkr-mcp yourself — your MCP client runs it for you based on the config below.

Configuration

Pick the client you use. The full setup guide, including a live picker that updates the snippet per client, lives at trakkr.ai/learn/api/mcp.

Claude Code (CLI)

Run once in any terminal:

claude mcp add trakkr \
  -e TRAKKR_API_KEY=sk_live_your_key_here \
  -- uvx trakkr-mcp

Claude Desktop

~/Library/Application Support/Claude/claude_desktop_config.json (macOS) or %APPDATA%\Claude\claude_desktop_config.json (Windows):

{
  "mcpServers": {
    "trakkr": {
      "command": "uvx",
      "args": ["trakkr-mcp"],
      "env": {
        "TRAKKR_API_KEY": "sk_live_your_key_here"
      }
    }
  }
}

Cursor

~/.cursor/mcp.json (global) or .cursor/mcp.json (project):

{
  "mcpServers": {
    "trakkr": {
      "command": "uvx",
      "args": ["trakkr-mcp"],
      "env": {
        "TRAKKR_API_KEY": "sk_live_your_key_here"
      }
    }
  }
}

VS Code (Copilot Chat)

.vscode/mcp.json — uses the input-prompt pattern so the key never lands in source:

{
  "servers": {
    "trakkr": {
      "type": "stdio",
      "command": "uvx",
      "args": ["trakkr-mcp"],
      "env": {
        "TRAKKR_API_KEY": "${input:trakkr-api-key}"
      }
    }
  },
  "inputs": [
    {
      "type": "promptString",
      "id": "trakkr-api-key",
      "description": "Trakkr API Key",
      "password": true
    }
  ]
}

Codex CLI

~/.codex/config.toml:

[mcp_servers.trakkr]
command = "uvx"
args = ["trakkr-mcp"]
env = { TRAKKR_API_KEY = "sk_live_your_key_here" }

Windsurf

~/.codeium/windsurf/mcp_config.json:

{
  "mcpServers": {
    "trakkr": {
      "command": "uvx",
      "args": ["trakkr-mcp"],
      "env": {
        "TRAKKR_API_KEY": "sk_live_your_key_here"
      }
    }
  }
}

Zed

~/.config/zed/settings.json:

{
  "context_servers": {
    "trakkr": {
      "source": "custom",
      "command": {
        "path": "uvx",
        "args": ["trakkr-mcp"],
        "env": {
          "TRAKKR_API_KEY": "sk_live_your_key_here"
        }
      }
    }
  }
}

Cline, Continue, or other MCP clients

Most clients accept the standard mcpServers JSON shown above for Cursor/Claude Desktop/Windsurf. Check your client's MCP documentation for the exact config file location, and see trakkr.ai/learn/api/mcp for an up-to-date picker.

Available Tools

Core Data

Tool	Description
list_brands	List all brands you're tracking. Returns brand IDs needed for other tools.
get_visibility_scores	Get AI visibility scores and trends over time.
list_prompts	List tracked search queries for a brand.
manage_prompt	Create, update, or delete a tracked prompt.

Visibility

Tool	Description
get_citations	Citation URLs and trends. Views: list, history, queries, sources, feed, heatmap.
get_rankings	Competitive rankings in AI search results.
get_model_breakdown	Visibility by AI model (ChatGPT, Perplexity, Gemini, etc.).
get_competitors	Competitor analysis. Views: summary, arena, head-to-head, by-model, threats.

Intelligence

Tool	Description
get_opportunities	Citation gaps where competitors appear but you don't.
get_content_ideas	AI-generated content ideas to improve visibility.
get_perception	How AI models describe and position the brand. Paid plan required.
get_prism	AI Pages analysis: positioning, strengths, weaknesses, opportunities. Paid plan required.
get_narratives	Narrative intelligence: tracked topics and storylines. Scale plan required.

Crawler

Tool	Description
get_crawler_overview	Overview tab: hero stats, chart, setup state, top pages, and recent preview.
get_crawler_live	Live tab: activity, pages, or sessions with dashboard filters.
get_crawler_pages	Pages tab: pages, paths, or normalized bots.
get_crawler_page_details	Page drawer data: verdict, pipeline, health, traffic, diagnostics.
get_crawler_path_details	Path drawer data: verdict, pipeline, top pages, top bots, diagnostics.
get_crawler_bot_details	Bot drawer data: verdict, pipeline, top pages, top paths, diagnostics.
get_crawler_access	Access tab: findings, bot matrix, robots.txt, llms.txt, submit-to-search.
preview_crawler_access_fix	Preview an Access fix without applying it.
send_crawler_verification_ping	Send the verification ping shown in the dashboard.
submit_crawler_to_search	Submit crawler pages to AI search via IndexNow.
get_crawler_submit_status	Read submit-to-search status and summary.
get_crawler_analytics	Deprecated compatibility alias for the new crawler tools.

Audit & Recommendations

Read-only access to Trakkr's unified recommendation queue and site-audit data. These tools never trigger analysis or generation — they surface data the dashboard has already produced. Ideal for running an AI-search technical audit conversationally.

Tool	Description
get_actions	List recommended actions from the unified queue (audit, crawler, prompts, citations, etc.). Filter by source, category, action_type, lens, or quick_win.
get_action_stats	Aggregate counts by status, category, action_type, and source.
list_audits	List site audits for a brand (overall_score, issue counts, detected CMS).
get_audit_findings	Audit issues + flagged pages for the latest (or specified) audit. Includes has_llms_txt / ai_crawler_blocked / has_sitemap rollups.
list_page_analyses	List recent cached deep page analyses for a brand.
get_page_analysis	Cached deep analysis for one URL — AI diagnosis, citation verdict, failed checks, ready-to-paste schema, entities, bot visibility.

Actions

Tool	Description
run_diagnosis	Diagnose a search query across AI models in real-time.
get_diagnosis_result	Get diagnosis results, history, or usage quota.
generate_report	Generate an AI visibility report (executive, weekly, or full).
get_reports	List or retrieve generated reports.
export_data	Export data as JSON or CSV.

Example Conversations

Once connected, you can ask your AI assistant things like:

• "How is my brand doing in AI search this month?"
• "Which competitors are gaining ground in ChatGPT?"
• "What content opportunities am I missing?"
• "Show me which pages get cited most by Perplexity"
• "Run a diagnosis on 'best project management tools'"
• "What does my visibility trend look like over the last 90 days?"
• "Show crawler overview for Nike"
• "What changed in Live this week?"
• "Preview crawler access fixes"
• "Send a verification ping"
• "Run a full AI search technical audit of my brand"
• "What are my top open actions right now?"
• "Show me every critical/high issue from the latest audit, and the worst-scoring pages"
• "What's the deep page analysis say about /pricing?"

Requirements

• Python 3.10+
• A Trakkr account with an API key (get one here)

Documentation

Full API documentation: trakkr.ai/learn/api

MCP setup guide: trakkr.ai/learn/api/mcp

License

MIT