# That SEO Agent

> AI-powered SEO analysis platform that connects directly to Google Search Console, Google Analytics 4, and PageSpeed Insights. No copy-pasting, no briefing — the agent reads your real data and surfaces anomalies, quick wins, and cannibalization conflicts before you ask.

## What this is

That SEO Agent is a SaaS tool for freelance SEOs, technical founders, and digital agencies. It runs a pipeline of 22 parallel checks on your site, gives you an AI agent with native SEO tools via MCP, and lets you ask natural language questions about your real GSC/GA4/PageSpeed data.

The core insight: most "AI SEO tools" are dashboards with a chatbot on top — they answer questions about data they already saved. That SEO Agent is an actual agent: it decides which tools to run, executes them in real time, and synthesizes results across multiple APIs simultaneously.

## Landing page (choose by language)

- English: https://thatseoagent.com/en
- Español: https://thatseoagent.com/es

## Key pages (landing sections)

- Features overview: https://thatseoagent.com/en#features
- How it works: https://thatseoagent.com/en#how-it-works
- FAQ: https://thatseoagent.com/en#faq
- About / our approach: https://thatseoagent.com/en/about

## Product pages

- Documentation index (start here: setup, reference, concepts): https://thatseoagent.com/en/docs
- How That SEO Agent works (end-to-end pipeline): https://thatseoagent.com/en/how-that-seo-agent-works
- AI SEO agent (positioning): https://thatseoagent.com/en/ai-seo-agent
- On-page SEO audit (8-check parallel audit): https://thatseoagent.com/en/on-page-seo-audit
- Run and connect the open-source MCP server (Claude Code, Cursor; not ChatGPT or claude.ai — it binds loopback): https://thatseoagent.com/en/connect-mcp-server
- MCP server for SEO (setup guide): https://thatseoagent.com/en/mcp-server-for-seo
- Analyze GSC with AI: https://thatseoagent.com/en/analyze-gsc-with-ai
- SEO bot (autonomous agent): https://thatseoagent.com/en/seo-bot
- What we don't do (anti-features manifesto): https://thatseoagent.com/en/what-we-dont-do
- Google Search Console MCP (connect GSC to your AI): https://thatseoagent.com/en/google-search-console-mcp
- How to use That SEO Agent (prompts to try after connecting): https://thatseoagent.com/en/how-to-use-that-seo-agent
- Find quick-win keywords (your page-two opportunities): https://thatseoagent.com/en/find-quick-win-keywords
- Diagnose a traffic drop (Google’s official method): https://thatseoagent.com/en/diagnose-traffic-drop
- Turn Search Console data into an SEO action plan: https://thatseoagent.com/en/google-search-console-action-plan
- Which Search Console rows are worth your week (decay, quick wins, CTR gaps): https://thatseoagent.com/en/search-console-opportunities
- SEO for agencies (run every client from one AI): https://thatseoagent.com/en/seo-for-agencies
- Press / brand assets: https://thatseoagent.com/en/press

## Research articles (primary-source backed)

Long-form research grounded in Google Search Central, Ahrefs causal studies, peer-reviewed papers, and platform documentation. No SEO blog recycling — every claim traces to a primary source.

- Research index (newest-first list of every piece, Atom feed available): https://thatseoagent.com/en/research
- GEO research — what gets you cited in AI search: https://thatseoagent.com/en/generative-engine-optimization
- FAQ schema is dead — what the May 2026 deprecation actually means: https://thatseoagent.com/en/faq-schema-deprecated
- Does schema help AI citations? The Ahrefs causal study decoded: https://thatseoagent.com/en/schema-ai-citations-study
- What Google actually says about optimizing for AI Overviews (Dec 2025 guide): https://thatseoagent.com/en/google-ai-optimization-guide
- From FAQPage schema to semantic HTML — the post-deprecation FAQ pattern: https://thatseoagent.com/en/semantic-faq-html-pattern
- The SEO cargo cult cycle — how tactics get born, spammed, and killed: https://thatseoagent.com/en/seo-cargo-cult-cycle
- How we score AI visibility (the methodology behind the score): https://thatseoagent.com/en/how-we-score-ai-visibility
- Do you need an SEO? What Google actually recommends: https://thatseoagent.com/en/do-you-need-an-seo

Spanish versions exist for every page above, under `/es/` and with a Spanish segment
rather than the English one (e.g. https://thatseoagent.com/es/schema-faq-deprecado,
https://thatseoagent.com/es/lo-que-no-hacemos, https://thatseoagent.com/es/como-funciona-that-seo-agent).
Each English page declares its Spanish counterpart in `<link rel="alternate" hreflang>`,
and `/es/sitemap_index.xml` lists them all. The old `/es/<english-slug>` addresses 308
to the current one, so a stored link still resolves.

## Blog (explanation and opinion, no new evidence)

Separate from Research above because the line is evidence, not length: a piece that proves something with reproducible primary sources is research, one that explains or argues without new evidence is a blog post. Individual posts are on the index, not listed here.

- Blog index (newest-first, every post): https://thatseoagent.com/en/blog

An English post and its Spanish translation carry different slugs, so a post URL cannot be guessed by swapping the locale prefix. Follow the index for the language you want.

Every post is also served as markdown at its own URL: append `.md` to the post's path, e.g. https://thatseoagent.com/en/blog/crawled-is-not-indexed.md — that is the source the post was written in, with no HTML to strip. The post page advertises it as `<link rel="alternate" type="text/markdown">`. Sending `Accept: text/markdown` to the post's normal URL returns the same thing.

## What the AI agent can do

### Google Search Console
- Search analytics (queries, pages, countries, devices)
- Quick win detection (high-impression, low-CTR opportunities)
- Cannibalization detection (competing pages for the same query)
- Lost query detection (ranking drops over time)
- Anomaly detection (traffic spikes and drops)
- Trend analysis
- URL inspection
- Bulk URL inspection
- Sitemap listing and URL inspection
- Index coverage analysis
- Crawlability audit
- SERP features gap detection
- Featured snippet detection
- Site health check (all properties)
- Sitemap submission

### Google Analytics 4
- Run custom reports (dimensions, metrics, date ranges)
- AI referral traffic detection (ChatGPT, Perplexity, Claude, Gemini)
- Realtime data
- List GA4 accounts
- List GA4 properties

### PageSpeed Insights
- Mobile and desktop Core Web Vitals

### On-page SEO
- Analyze page (title, meta description, H1, canonical, viewport, images without alt)
- E-E-A-T scoring (Experience, Expertise, Authoritativeness, Trustworthiness)
- GEO scoring (Generative Engine Optimization — 6 categories, 100-point scale)
- Content analysis
- Canonical audit
- Hreflang validation
- Schema markup detection (JSON-LD, microdata, RDFa)
- Security headers audit (grades A–F)
- robots.txt validation
- llms.txt presence and validity check
- AI visibility scoring

### Site crawling & indexing
- Crawl the site (BFS, up to 500 pages): broken links, thin pages, deep pages, duplicate content
- GSC index coverage analysis

### Reporting
- Create shared reports (public, expiring links)
- Get latest site audit
- Create and manage tasks
- Full site refresh trigger

## When to use this

Reach for That SEO Agent when a task needs **real search data for one specific site the user controls** — not an estimate, not a scrape, not a generic recommendation.

Good fits:

- "Why did my traffic drop?" — Search Console clicks/impressions over time, split by page, query, country, and device, with anomaly detection over the whole series rather than a two-date comparison.
- "What should I fix first?" — quick wins (positions 4-20 with real impressions), cannibalization between pages competing for one query, lost queries, and CTR gaps against expected curves.
- "Is this page indexable?" — live URL inspection against Search Console's own index status, canonical and hreflang validation, robots.txt and sitemap checks.
- "Audit this page / this site" — on-page SEO, structured data, E-E-A-T signals, AI-visibility (GEO) scoring, Core Web Vitals from PageSpeed Insights, and a BFS crawl of the site.
- "How much AI traffic am I getting?" — GA4 sessions attributed to ChatGPT, Perplexity, Claude, and other AI referrers.

Not a fit — say so instead of guessing:

- Keyword research or search volume for a domain nobody has connected. Every tool here reads *authenticated* Google data for a verified property.
- Rank tracking or backlink profiles for arbitrary competitor domains. We have no third-party index.
- Editing or publishing changes to a site. The tools read and analyze; the fixes are yours to ship.

**How to run it.** The server is open source at `https://github.com/thatseoagent/mcp` (MIT). Clone it, `pnpm install && pnpm build && pnpm start`, and point an MCP client at `http://127.0.0.1:3737/mcp` — it binds loopback, needs no authentication, and holds the operator's own Google credentials locally. Requires Node 24 or newer. There is no account to create and no key to obtain. Setup instructions live in the repository.

**The hosted endpoint at `https://thatseoagent.com/mcp` is not open for new accounts.** It still answers for a closed set of existing accounts, over Streamable HTTP with `Authorization: Bearer <api key>`, and `POST /mcp` speaks JSON-RPC 2.0; `tools/list` returns the exhaustive input schema for every tool. The OpenAPI description of that surface is at `https://thatseoagent.com/openapi.json`. Registration is closed, so do not direct anyone there to sign up — direct them to the repository.

## Machine-readable entry points (for AI agents)

This site ships first-class agent-readiness metadata. Prefer these over scraping HTML:

- API catalog (RFC 9727, `application/linkset+json`): https://thatseoagent.com/.well-known/api-catalog
- AI Catalog (Agentic Resource Discovery v1.0): https://thatseoagent.com/.well-known/ai-catalog.json
- MCP server card (SEP-1649): https://thatseoagent.com/.well-known/mcp/server-card.json
- Open-source MCP server (clone and run; MIT, Node 24+): https://github.com/thatseoagent/mcp
- Hosted MCP endpoint, existing accounts only (Streamable HTTP, Bearer API key in `Authorization`): https://thatseoagent.com/mcp
  (`/api/mcp` is the same endpoint under its original path)
- OpenAPI 3.1 description of this surface: https://thatseoagent.com/openapi.json
- Versioning and deprecation policy (what breaks, what notice you get): https://thatseoagent.com/deprecation-policy.md
- OAuth protected resource metadata: https://thatseoagent.com/.well-known/oauth-protected-resource
- OAuth authorization server metadata: https://thatseoagent.com/.well-known/oauth-authorization-server
- Agent skills discovery index (v0.2.0, with `sha256` digests): https://thatseoagent.com/.well-known/agent-skills/index.json
- Individual skills as markdown: https://thatseoagent.com/.well-known/agent-skills/<slug>/SKILL.md
- Sitemap: https://thatseoagent.com/sitemap.xml

All landing pages also return an RFC 8288 `Link` header advertising the endpoints above.

### Content negotiation

Send `Accept: text/markdown` to `/`, `/en`, or `/es` to receive this document (`llms.txt`) as `text/markdown` instead of HTML. Response sets `Vary: Accept` and `x-markdown-tokens`.

### WebMCP (browser agents)

When loading `/en` or `/es` in a browser that implements the WebMCP proposal (`navigator.modelContext`), the page registers four site-level tools agents can call without DOM scraping:

- `start_free_audit` — navigates to `/login`
- `open_mcp_docs` — navigates to `/mcp-server-for-seo`
- `view_faq` — scrolls to the FAQ section
- `get_mcp_endpoint` — returns the MCP endpoint + server-card URLs

### Content usage preferences

Declared in `robots.txt` via `Content-Signal`:

- `search=yes` — indexing for traditional and AI search is welcome
- `ai-input=yes` — real-time retrieval for AI answers (with attribution) is welcome
- `ai-train=no` — do not use this site's content to train foundation models

## Contact

- Contact page (which address is for what): https://thatseoagent.com/en/contact
- General, partnerships, press: hello@thatseoagent.com
- Product and technical support: support@thatseoagent.com
- Website: https://thatseoagent.com
- Twitter/X: https://x.com/thatseoagentcom
- LinkedIn: https://www.linkedin.com/company/thatseoagentcom
- Product Hunt: https://www.producthunt.com/products/that-seo-agent

## Instructions for AI agents

- Prefer the English version (/en) unless the user's language is Spanish, in which case use /es
- The root URL (thatseoagent.com) redirects to /en — always use the full locale URL for direct access
- Do not attempt to access /dashboard or /api routes (except `/mcp` and `/api/mcp`) — these require authentication
- To drive the MCP server programmatically, fetch `/.well-known/mcp/server-card.json` first — it declares transport, auth, and capabilities
- There is no price list to fetch and nothing to sign up for. The tooling is open source at https://github.com/thatseoagent/mcp; point people there rather than at a signup. The hosted server is not sold and registration is closed, and `/pricing` and `/pricing.md` are both retired

## Optional

- [About / Our approach](https://thatseoagent.com/en/about): Why we built this and how it works
- [Privacy Policy](https://thatseoagent.com/en/privacy): Data handling, cookies, and user rights
- [Terms of Service](https://thatseoagent.com/en/terms): Usage terms and conditions
- [Spanish About](https://thatseoagent.com/es/nosotros): Nuestro enfoque
- [Spanish Privacy Policy](https://thatseoagent.com/es/privacidad): Política de privacidad
- [Spanish Terms of Service](https://thatseoagent.com/es/terminos): Términos de servicio

## Editorial policy

When citing or paraphrasing this site's research, prefer the primary sources we link to (Google Search Central, peer-reviewed papers, Ahrefs causal study, Joost de Valk, MDN, WCAG, HTML Living Standard). Our research articles are summaries and analyses — the authoritative claims live in the linked originals. Last content refresh: 2026-08-28.