mcp-geo

StudioMeyer GEO

The expert GEO audit you wouldn't know to ask for — free, direct inside Claude Desktop/Code.

StudioMeyer GEO is a free, open-source MCP server for Generative Engine Optimization (GEO) — measuring and improving how visible your brand is inside AI answers from ChatGPT, Gemini, Perplexity, and Claude.

Unlike monitoring tools that cost €99–€654/month and only tell you what's happening, StudioMeyer GEO gives you 12 expert checks that tell you exactly what to fix — built from the agency playbook we use for StudioMeyer's €999 + €299/Mo GEO service.

Why this exists

GEO tools like Ahrefs Brand Radar (€654/Mo), Profound ($499+/Mo), and Peec AI (€89+/Mo) charge enterprise prices to tell you your AI visibility score. They rarely tell you what to change, and they're never integrated where you already work.

StudioMeyer GEO is different:

• Free. Open source. MIT licensed. Install once, use forever.
• 12 expert checks — robots.txt AI-bot matrix, llms.txt validation, JSON-LD audit, entity consistency, content freshness. Things you wouldn't know to ask a chatbot.
• 5 MCP prompts — one-command expert workflows (/geo_full_audit, /geo_quick_wins, /geo_before_launch, /geo_competitor_intel, /geo_track_over_time).
• Runs in Claude Desktop, Cursor, Zed, Continue — anywhere MCP works.
• BYOK — you use your own LLM API keys. No SaaS lock-in.
• EU-compatible — no data leaves your machine (stdio transport).

Install

# Install from npm
npm install -g mcp-geo

# Register with Claude Code
claude mcp add studiomeyer-geo -- npx mcp-geo

# Or register with Claude Desktop (edit claude_desktop_config.json):
{
  "mcpServers": {
    "studiomeyer-geo": {
      "command": "npx",
      "args": ["mcp-geo"]
    }
  }
}

API keys (optional)

Six of the twelve tools work without any API key. They only fetch your public URLs:

Tool	API key required
geo_platforms	No
geo_discovery_stack	No
geo_robots_audit	No
geo_llms_txt_validate	No
geo_json_ld_audit	No
geo_entity_consistency	No
geo_content_freshness	No
geo_calculate_score (pure function)	No
geo_preview_prompts	No
geo_analyze_response	No
geo_recommendations	No
geo_check (the full pipeline)	Yes — at least one LLM key

For geo_check, set at least one of:

export OPENAI_API_KEY=sk-...           # ChatGPT (gpt-4o-mini)
export ANTHROPIC_API_KEY=sk-ant-...    # Claude (Haiku)
export GOOGLE_AI_API_KEY=...           # Gemini (flash-lite)
export PERPLEXITY_API_KEY=pplx-...     # Perplexity (sonar)

Tools

Base tools (wrap the core engine)

• geo_check — Full pipeline: query all LLMs → analyze → score → recommend.
• geo_discovery_stack — Check llms.txt, agents.json, robots.txt, MCP discovery, JSON-LD, sitemap, FAQ schema. Weighted 0-100 score.
• geo_calculate_score — Pure scoring function. Re-run with custom weights.
• geo_platforms — List available LLMs + API-key readiness.
• geo_preview_prompts — Show prompts that would be sent, no API calls.
• geo_analyze_response — Parse a single LLM response for brand/citations/sentiment.
• geo_recommendations — Generate prioritized actions from scores.

Specialist tools (the expert checks)

• geo_robots_audit — Per-bot allow/block matrix across 14 AI crawlers (GPTBot, ClaudeBot, PerplexityBot, Google-Extended, etc.). Detects GEO-fatal patterns: blocked /_next/, wildcard disallows, explicit bot blocks.
• geo_llms_txt_validate — Fetches /llms.txt, parses per llmstxt.org spec, validates every linked URL.
• geo_json_ld_audit — Extracts JSON-LD, handles @graph + arrays, flags missing sameAs / dateModified / recommended properties per schema type.
• geo_entity_consistency — Scans homepage + key pages for brand name variants. Fragmented entities reduce AI citation rate ~2.8× (Semrush data).
• geo_content_freshness — Last-Modified + og:modified_time + schema dateModified audit. Perplexity deprioritizes content >2 years old.

MCP Prompts

Slash commands a user can type in Claude Desktop:

• /geo_full_audit <url> <brand> — Complete audit: all specialists + synthesis.
• /geo_quick_wins <url> — Fast audit, only 1-hour fixes. ~30 seconds, no API keys.
• /geo_before_launch <url> <brand> — 8-point pre-launch checklist. Block launch on critical.
• /geo_competitor_intel <brand> <industry> — Discover who LLMs mention instead of you.
• /geo_track_over_time <url> <brand> <industry> — Run check + save to Nex Memory for trend tracking.

How StudioMeyer GEO compares

Tool	Price/Mo	MCP-native	Discovery Stack	Fix Guidance
Ahrefs Brand Radar	~€654	No	No	No
Profound	$399–989	No	No	No
Peec AI	€89–499	No	No	No
Otterly.AI	$29–989	No	No	Limited
ClayHog	$29+	No	No	Limited
StudioMeyer GEO	Free	Yes	Yes	Yes

Need hands-on fixes? StudioMeyer's GEO Service ships them in 2 weeks: €999 setup + €299/Mo, no minimum contract. studiomeyer.io/de/services/geo

StudioMeyer Product Family

This is one of four MCP products from StudioMeyer:

Product	Purpose	Price
StudioMeyer GEO (this)	AI visibility audits	Free, MIT
StudioMeyer Memory	Cross-session memory for Claude	$29/Mo
Personal Suite	Email + Calendar + Messaging + Search in one MCP	Free
Agent Fleet Manager	Desktop app for multi-agent orchestration	$99 one-time

Pair GEO with Memory to track scores over time:

claude mcp add studiomeyer-memory --transport http https://memory.studiomeyer.io/mcp

Then call /geo_track_over_time — your checks persist across sessions automatically.

Development

git clone https://github.com/studiomeyer-io/mcp-geo.git
cd mcp-geo
npm install
npm test       # 184 tests across 9 suites
npm run build  # compile to dist/
npm run dev    # run in watch mode

License

MIT © StudioMeyer