Skip to main content
v1.0GuidesGuidesAPI ReferenceAPI Reference

Agent Analytics Server

Analyze how AI bots and crawlers interact with your site: total visits, AI traffic share, pages indexed, referrals from AI search, per-platform breakdowns, trends over time, and AI-content SEO health.

Overview​

The Agent Analytics Server exposes data about how AI bots/crawlers (ChatGPT, Perplexity, Anthropic, Google AI, etc.) visit, index, and refer traffic to your organization's site. It covers the four screens of the Agent Analytics product: Overview, Pages, Referrals, and AI Crawlers.

Server Details​

  • Server Name: agent-analytics-server
  • Version: 0.0.1
  • Endpoint: /mcp/agent-analytics
  • Authentication: Organization-level required (x-api-key header)

Purpose​

Answer questions about AI-crawler traffic and indexing: total visits, traffic share, pages indexed, and referrals for a date range; per-AI-platform breakdowns of the same metrics; individual page-level indexing and referral stats; trends over time; and AI-content SEO health (crawl runs and per-page optimization suggestions).

Data Model​

Tools are grouped by the four screens of the Agent Analytics product:

  • Overview screen: top-line totals for a date range, including All Visits, AI Traffic Percentage, Number of Pages Indexed, Referrals from AI Search, and an Indexing Analysis Breakdown of those same metrics per AI platform/bot.
  • Pages screen: indexing detail, including a per-platform Page Platforms Breakdown, a Pages Indexed Over Time trend, and a Pages List of individual pages with their own indexing/visit/referral stats.
  • Referrals screen: the same shape as Pages, but focused on the human visitors AI referred to your site rather than on indexing: Referrals Over Time, a Referral Platforms Breakdown, and a Recent Visits page list sorted by recency.
  • AI Crawlers screen: AI content and SEO health, including a Crawl Runs list (one per domain/site scanned) and, per crawl run, a Crawl Run Pages list with AI-content optimization suggestions (missing canonical URL, meta description, Open Graph tags, and so on).

An AI platform/bot is a named crawler or agent (OpenAI, Anthropic, Perplexity, Google AI, Apple AI, Amazon AI, Meta, DuckDuckGo AI, etc.). The same set appears in every per-platform breakdown tool across the first three screens.

  1. For a single overview number (visits, traffic %, pages indexed, referrals), call the matching Overview metric tool.
  2. For "which AI platform" or per-platform comparison questions, call Indexing Analysis Breakdown (Overview), Page Platforms Breakdown (Pages), or Referral Platforms Breakdown (Referrals), picking the one that matches the screen or metric the user means.
  3. For "which pages" or per-page questions, call Pages List (indexing-focused fields) or Recent Visits (referral-focused fields, sorted by recency).
  4. For trend, spike, or drop-over-time questions, call the matching "over time" tool: Pages Indexed Over Time or Referrals Over Time.
  5. For AI content and crawler questions ("is my site AI-crawler friendly", "what SEO issues does AI see"), call Crawl Runs List first to get a crawlRunId, then Crawl Run Pages.

Most tools return a formatted Markdown table with a one-line explanatory note underneath, so responses render directly in the assistant's answer without extra formatting work. The two "over time" tools are the exception: they return raw per-platform daily time-series data, since a date × platform matrix doesn't flatten well into a table.

Available Tools​

1. all-visits-tool​

Purpose: Total AI bot/crawler visits in a date range (Overview screen).

Input: startDate, endDate (both required).

Returns: A single number.

2. ai-traffic-percentage-tool​

Purpose: Percentage of total site traffic that comes from AI bots/crawlers (Overview screen).

Input: startDate, endDate (both required).

Returns: A single number.

3. number-of-pages-indexed-tool​

Purpose: Total count of unique pages indexed by AI bots/crawlers (Overview screen).

Input: startDate, endDate (both required).

Returns: A single number.

4. referrals-from-ai-search-tool​

Purpose: Count of human visitors referred to your site from an AI platform's answer (Overview screen).

Input: startDate, endDate (both required).

Returns: A single number.

5. indexing-analysis-breakdown-tool​

Purpose: Visits, referral visits, pages indexed, and traffic % change, broken down per AI platform/bot (Overview screen).

Input: startDate, endDate (both required).

Returns: One row per AI platform/bot.

6. page-platforms-breakdown-tool​

Purpose: Pages-indexed count and % change per AI platform/bot (the Pages screen's narrower platform rollup).

Input: startDate, endDate (both required).

Returns: One row per AI platform/bot.

7. pages-indexed-over-time-tool​

Purpose: Daily time series of pages indexed, per AI platform/bot (Pages screen trend chart).

Input: startDate, endDate (both required).

Returns: Raw per-platform daily series data, not a table.

8. pages-list-tool​

Purpose: Individual pages with their own indexing/visit/referral stats (Pages screen).

Input:

  • startDate, endDate (required)
  • search, orderField/orderDir, field, limit/offset (optional)

Returns: Page path, total indexed by AI, AI indexing, AI training indexing, people referred, % of bot referrals, last crawled date.

9. referrals-over-time-tool​

Purpose: Daily time series of AI-driven referral visits, per AI platform/bot (Referrals screen trend chart).

Input: startDate, endDate (both required).

Returns: Raw per-platform daily series data, not a table.

10. referral-platforms-breakdown-tool​

Purpose: Referral-visit count and % change per AI platform/bot (the Referrals screen's platform rollup).

Input: startDate, endDate (both required).

Returns: One row per AI platform/bot.

11. recent-visits-tool​

Purpose: Individual pages recently visited by AI bots, with referral stats (Referrals screen).

Input:

  • startDate, endDate (required)
  • search, orderField/orderDir, field, limit/offset (optional)

Returns: Page path, referral visits, % of bot referrals, total indexed by AI, last crawled date. Sorted by last-visited date by default.

note

This tool lists the same underlying page-level data as pages-list-tool, just with a different default sort and focus field: recency and referral_visits here, versus indexing and ai_visits there. Use whichever matches the screen the user is asking about.

12. crawl-runs-list-tool​

Purpose: List AI crawl runs, one per domain/site scanned for AI content analysis (AI Crawlers screen).

Input: search, limit/offset (optional).

Returns: Crawl run id, domain, status, total/processed/generated/failed URL counts, started and completed dates.

When to use: Call this first. Its id feeds crawl-run-pages-tool.

13. crawl-run-pages-tool​

Purpose: List pages within a crawl run, with AI-content optimization suggestions (AI Crawlers screen).

Input:

  • crawlRunId (required, from crawl-runs-list-tool)
  • search, limit/offset (optional)

Returns: Page path, processing status, token counts, and AI-content suggestions (for example, missing canonical URL, meta description, or Open Graph tags).

note

Suggestion names are abbreviated for readability. The underlying data also has a full description, issue, and fix per suggestion.

Notes​

  • All data is automatically scoped to the authenticated organization.
  • Every date-range tool requires startDate/endDate in YYYY-MM-DD format. The AI Crawlers screen tools are not date-range scoped.
  • Free-text search params use contains-match by default.

Connecting a Client​

The Agent Analytics Server is a remote HTTP MCP server. See the Claude Integration guide to connect it to Claude Code, Claude Desktop, or claude.ai, or the LLM Integration Guide for OpenAI.

Last updated on by Sona