GEO settings — webhouse.app Docs

AI visibility controls — which AI crawlers can index your site, what the robots.txt strategy is, and how your content surfaces in ChatGPT / Claude / Perplexity answers.

Where it is

Settings → GEO (/admin/settings?tab=geo).

GEO (Generative Engine Optimization) is about making your content discoverable and citable by AI platforms like ChatGPT, Claude, and Perplexity — similar to how SEO is for Google. This tab controls which crawlers can access your content and how the discovery files are generated.

Robots.txt strategy

The core control. Pick one of four strategies:

Strategy	What it does
Maximum	All bots allowed, including training bots. Best for maximum visibility. Default
Balanced	Search bots allowed (ChatGPT-User, Claude-SearchBot, PerplexityBot). Training bots blocked (GPTBot, ClaudeBot, Google-Extended). Your content powers AI answers but isn't used for model training
Restrictive	All AI bots blocked. Only traditional search engines allowed. Use if legally required
Custom	Define your own rules line by line. The tab exposes the generated `robots.txt` for review

The generated robots.txt also disallows /admin/, /api/, and preview paths on every strategy.

Generated discovery files

Every build writes these files at the site root (they're what crawlers and AI platforms look for):

robots.txt — crawler policy based on your strategy
sitemap.xml — all indexable pages, with <xhtml:link rel="alternate" hreflang> for multi-locale sites
llms.txt — AI-friendly index of your site structure + MCP endpoint pointer
llms-full.txt — full markdown dump of every published document, for retrieval-augmented AI use
feed.xml — RSS feed of your posts
ai-plugin.json — MCP plugin manifest so AI platforms can discover your live content API
Per-page .md files — every HTML page has a markdown sibling at the same URL + .md (e.g. /blog/post.html → /blog/post.md)

All of this is automatic once the deploy runs. The GEO tab lets you configure what goes in each, not the build mechanics.

Country / region restrictions

Optional. Most sites leave this blank. Set here if your content is geographically restricted (legal, regional licensing). The value feeds into:

<meta name="geo.region"> in every page
The llms.txt preamble
Structured data (areaServed) in JSON-LD for businesses

AI citation tuning

GEO scoring (shown in the dashboard) measures how well your content is structured for AI citation. This tab surfaces the rules and lets you override some:

Answer-first — lead paragraphs answer the H1 question directly
Question headers — H2s match how people actually ask
Statistics — include numbers, percentages, data points
Citations — link to authoritative external sources
Freshness — content updated within 90 days
JSON-LD — structured data for articles, FAQs, HowTo
Named author — E-E-A-T trust signals
Depth — 800+ words for comprehensive coverage

The tab shows which rules your site currently passes + the Optimize All button to run all of them against every published doc.

The mcp-plugin endpoint

When GEO is enabled, the CMS exposes a public MCP endpoint at /api/mcp/public. AI platforms that support MCP (Claude Desktop, Cursor, custom agents) can plug it in and get live, structured access to your content — not just the scraped HTML.