Innotek is the only GEO platform with native Model Context Protocol integration — for both Claude Desktop and Claude.ai web. Query live audit data, analyse competitors, and generate production-ready fixes — without leaving your conversation.
Model Context Protocol (MCP) is an open standard introduced by Anthropic in 2024 that allows AI agents to connect to external data sources and tools via standardised interfaces. Instead of switching between a dashboard and a conversation, MCP lets Claude call your live GEO data in real time.
Innotek's MCP server exposes 8 production tools that Enterprise plan users can access from both Claude Desktop and Claude.ai web — via a native OAuth 2.1 PKCE flow that authorises with your Innotek API key. No competitor — not Frase.io, not WordLift, not Semrush, not Surfer SEO — offers this capability. It's the difference between reading a report and having a conversation with your data.
Each tool connects to Innotek's live API and returns structured data Claude can reason about.
Fetches the most recent completed GEO audit for any URL. Returns the overall grade (A–F), Entity Clarity score (1–10), Fact Density score, Schema Completeness (0–100%), Premium AI Score (0–100), full JSON-LD structured data, and the optimised llms.txt entry. Average response time: under 3 seconds.
Fetches premium competitor research powered by Gemini with live Google Search grounding. Returns an AI-generated research summary, technical SEO notes identifying content gaps, competitor landscape analysis, and an optimised llms.txt entry. Requires a completed GEO audit first.
Generates production-ready HTML to fix a specific SEO issue identified in a GEO audit. Accepts the analysis ID, the specific issue description, and additional context. Returns clean HTML — Schema.org JSON-LD blocks, content sections, stat bars — ready to paste into your site. Consumes 1 analysis credit per generation.
Generates a complete prioritised Grade A remediation strategy for a site based on its most recent GEO audit. Pulls entity clarity scores, fact density, schema completeness, premium metrics, audit findings, and worst-performing pages. Returns a structured markdown plan with per-page action items, JSON-LD template, and a competitor context section if premium research has been run. Consumes 1 analysis credit.
Generates a GEO-informed SEO title tag and meta description for a specific crawled page. Uses the site’s entity clarity score, page-level audit findings, and llms.txt context to write copy that directly addresses what’s failing. No platform credits consumed — uses your connected OpenAI key.
Generates accurate Schema.org JSON-LD structured data for a specific crawled page. Selects the most specific applicable type (Article, FAQPage, LocalBusiness, Product, Service, HowTo…) based on actual page content. Validates @context and @type before returning. No platform credits consumed — uses your connected OpenAI key.
Rewrites any crawled landing page for a specific city. Injects at least 3 localised entities — landmarks, districts, transport links — naturally into the copy. Preserves brand name, pricing, and core offering. Returns pure Markdown matching the source structure. No platform credits consumed — uses your connected Gemini key.
Converts a competitor content gap (from get_competitor_research) into a ~2000-word authoritative blog post. Two-pass generation: Pass 1 produces high-density Markdown with H2/H3 structure and a data table; Pass 2 wraps it in a dark-themed branded HTML document. Returns full Markdown; HTML preview in conversation — use the REST API for the complete HTML. No platform credits consumed — uses your connected OpenAI or Gemini key.
Why MCP changes the workflow for SEO teams.
| Capability | Traditional Dashboard | Innotek MCP |
|---|---|---|
| Access method | Login → navigate → click → read | Ask Claude in natural language |
| Multi-site analysis | One tab per site, manual comparison | "Compare these 3 sites" in one message |
| Fix generation | Read recommendation → write code manually | Claude generates production HTML inline |
| Competitor research | Separate tool, separate login | Same conversation, same context |
| Team sharing | Screenshot or export PDF | Share the conversation thread |
| Learning curve | Learn the UI, memorise navigation | Just describe what you need |
Subscribe to the Innotek Enterprise plan (£299/mo). MCP access is included — your API key from Settings → Developer → API Keys is your credential.
Open Claude.ai → Settings → Integrations → Add MCP Server. Paste your Innotek server URL. You'll be redirected to an authorisation page — enter your Enterprise API key and click "Grant Access". No desktop app required.
Install Claude Desktop from claude.ai/download. Add the Innotek MCP server to your configuration file using the snippet provided in Settings → Developer → API Keys. The server URL, API key, and a ready-to-paste JSON block are shown there.
Start a new conversation in Claude (web or desktop). All 8 Innotek tools will appear in Claude's available tools. Ask any question — "audit my site", "compare me to frase.io", "localise my homepage for Manchester" — and Claude will call the appropriate tool automatically.