MCP Server

GEO Optimizer exposes its core functionality as an MCP (Model Context Protocol) server, making all tools available inside AI coding assistants like Claude Code, Cursor, and Windsurf.

Installation

Install with the mcp extra:

pip install geo-optimizer-skill[mcp]

This installs the geo-mcp entry point alongside the mcp Python SDK.

Setup

Claude Code

claude mcp add geo-optimizer -- geo-mcp

That’s it. The tools are immediately available in your Claude Code session.

Cursor

Add to your .cursor/mcp.json:

{
  "mcpServers": {
    "geo-optimizer": {
      "command": "geo-mcp",
      "args": []
    }
  }
}

Windsurf / Generic MCP Client

Any MCP client that supports stdio transport can connect:

geo-mcp

The server runs on stdio transport by default. No port configuration needed.

Direct Python Invocation

python -m geo_optimizer.mcp.server

Tools (10)

1. geo_audit

Run a complete GEO audit on a website. Analyzes 7 areas: robots.txt, llms.txt, JSON-LD schema, SEO meta tags, content quality, signals, and AI discovery. Returns a score 0–100 with detailed breakdown and recommendations.

Example output (abbreviated):

{
  "url": "https://example.com",
  "score": 72,
  "band": "good",
  "score_breakdown": {
    "robots": 18,
    "llms": 6,
    "schema": 16,
    "meta": 14,
    "content": 10,
    "signals": 5,
    "ai_discovery": 3
  },
  "recommendations": [
    "Add llms-full.txt for comprehensive AI indexing",
    "Add FAQPage schema for better AI extraction"
  ]
}

2. geo_fix

Generate automatic GEO fixes for a website. Audits the site and produces corrective artifacts: robots.txt patches, llms.txt content, missing JSON-LD schemas, and HTML meta tag snippets.

Example output (abbreviated):

{
  "fixes": [
    {
      "category": "robots",
      "description": "Create robots.txt with access for all 27 AI bots",
      "file_name": "robots.txt",
      "action": "create"
    },
    {
      "category": "llms",
      "description": "Generate llms.txt from sitemap (42 URLs)",
      "file_name": "llms.txt",
      "action": "create"
    }
  ],
  "score_before": 34,
  "score_estimated_after": 78
}

3. geo_llms_generate

Generate complete llms.txt content for a website. Discovers the sitemap, categorizes URLs by content type, and produces a structured markdown file following the llmstxt.org specification.

Example output:

# Example Site

> A brief description of the site for AI search engines.

## Documentation

- [Getting Started](https://example.com/docs/getting-started)
- [API Reference](https://example.com/docs/api)

## Blog & Articles

- [How We Built X](https://example.com/blog/how-we-built-x)

4. geo_citability

Analyze content citability using the 47 GEO methods implemented in GEO Optimizer. Evaluates the page content and returns a citability score 0–100 with per-method breakdown.

Example output (abbreviated):

{
  "url": "https://example.com/blog/post",
  "score": 65,
  "methods": {
    "quotation_addition": { "score": 8, "max": 10 },
    "statistics_addition": { "score": 10, "max": 12 },
    "fluency_optimization": { "score": 9, "max": 12 },
    "cite_sources": { "score": 6, "max": 10 },
    "answer_first": { "score": 7, "max": 10 },
    "passage_density": { "score": 5, "max": 10 }
  },
  "suggestions": [
    "Add 2-3 direct quotes from domain experts",
    "Add inline citations to authoritative sources"
  ]
}

5. geo_schema_validate

Validate a JSON-LD schema against schema.org requirements. Checks that all required fields are present for the given schema type.

Example output:

{
  "valid": true,
  "error": null,
  "schema_type": "website"
}

6. geo_compare

Compare GEO scores across multiple websites (max 5). Returns a ranked comparison with score, band, and per-category breakdown.

Example output:

{
  "comparison": [
    { "url": "https://site1.com", "score": 82, "band": "good" },
    { "url": "https://site2.com", "score": 45, "band": "foundation" }
  ],
  "total_sites": 2
}

7. geo_gap_analysis

Interpret the GEO gap between two sites. Identifies the weaker site, estimates the score delta, and returns prioritized actions with estimated point impact and CLI commands where available.

Example output:

{
  "weaker_url": "https://mysite.com",
  "stronger_url": "https://competitor.com",
  "score_gap": 23,
  "action_plan": [
    {
      "category": "llms",
      "title": "Publish llms.txt",
      "impact_points": 5,
      "command": "geo llms --base-url https://mysite.com"
    }
  ]
}

8. geo_ai_discovery

Check AI discovery endpoints on a website. Verifies the presence of /.well-known/ai.txt, /ai/summary.json, /ai/faq.json, and /ai/service.json based on the geo-checklist.dev emerging standard.

Example output:

{
  "well_known_ai_txt": true,
  "ai_summary_json": true,
  "ai_faq_json": false,
  "ai_service_json": false,
  "score": 4,
  "max_score": 6
}

9. geo_check_bots

Check which AI bots can access a website via robots.txt. Returns per-bot status (allowed/blocked/missing) with 3-tier classification (training/search/user) and citation bot verification.

Example output (abbreviated):

{
  "url": "https://example.com",
  "robots_found": true,
  "citation_bots_ok": true,
  "bots": {
    "OAI-SearchBot": { "description": "OpenAI (ChatGPT search citations)", "status": "allowed", "tier": "search" },
    "ClaudeBot": { "description": "Anthropic (Claude citations)", "status": "allowed", "tier": "search" },
    "GPTBot": { "description": "OpenAI (ChatGPT training)", "status": "blocked", "tier": "training" }
  },
  "summary": { "allowed": 20, "blocked": 2, "missing": 2 }
}

10. geo_trust_score

Return the Trust Stack Score for a website. Aggregates five trust layers: technical, identity, social, academic, and consistency.

Example output (abbreviated):

{
  "url": "https://example.com",
  "composite_score": 18,
  "grade": "B",
  "trust_level": "high"
}

11. geo_negative_signals

Check negative signals that reduce AI citation probability. Detects CTA overload, thin content, keyword stuffing, mixed signals, and other anti-citation patterns.

Example output (abbreviated):

{
  "checked": true,
  "severity": "medium",
  "signals_found": 2,
  "cta_count": 4,
  "has_keyword_stuffing": false
}

12. geo_factual_accuracy

Audit factual claims, sourcing quality, and obvious contradictions on a page. The tool flags unsourced numeric or evidence-style claims, surfaces unverifiable wording, highlights date/number inconsistencies, and checks linked sources for obvious failures.

Example output:

{
  "checked": true,
  "claims_found": 4,
  "claims_sourced": 2,
  "claims_unsourced": 1,
  "unsourced_claims": [
    "Studies show 42% of users prefer GEO workflows."
  ],
  "inconsistencies": [
    "Conflicting numeric claims for 'conversion rate': 42%, 45%"
  ],
  "broken_source_links": [
    "https://broken.example.com/report"
  ],
  "severity": "high"
}

Resources (5)

MCP resources provide read-only reference data. Access them via the geo:// URI scheme.

Usage Examples

Claude Code — Full Site Audit

> Use geo_audit to check https://mysite.com and tell me what to fix first.

Claude will call the geo_audit tool, parse the results, and give you prioritized recommendations in natural language.

Claude Code — Generate and Validate Schema

> Generate a WebSite JSON-LD schema for mysite.com, then validate it with geo_schema_validate.

Cursor — Compare Competitors

> Use geo_compare to compare mysite.com against competitor1.com and competitor2.com.
> Show me where I'm losing points.

Fix Workflow

> Run geo_fix on https://mysite.com with only=robots,llms.
> Show me the generated files.

Security

All URL inputs are validated against SSRF attacks before making any HTTP requests. Private IPs, localhost, and internal network addresses are rejected. HTTP response size is capped at 10 MB.