Innotek Solutions Innotek Solutions
← Back to Blog

· Innotek SEO Team

Using InnotekSEO AI Inside Claude with MCP

IS

Innotek SEO Team

Enterprise Software & GEO Consultants at Innotek Solutions Ltd — 16+ years of Microsoft and AI-powered search expertise.

MCPClaudeGEOIntegration

The Model Context Protocol (MCP) is an open standard that lets AI assistants connect to external tools and data sources. InnotekSEO AI provides a native MCP server, which means you can run GEO audits, generate HTML fixes, research competitors, and optimise content — all without leaving Claude.

This integration represents a fundamental shift in how SEO and GEO workflows operate. Instead of switching between a dashboard, a spreadsheet, and your CMS, you can audit, analyse, and fix issues in a single conversational flow. The AI assistant becomes your GEO co-pilot, calling the right tools at the right time and presenting results in context.

What's Available?

InnotekSEO AI exposes 8 tools via MCP, divided into two categories:

Core Audit Tools

These tools are available on all plans, including Free:

  • get_geo_audit — Run a complete GEO audit on any URL. Returns Entity Clarity, Fact Density, Schema Completeness, GEO Grade, and AI Readiness scores with detailed breakdowns across all dimensions.
  • get_competitor_research — Gemini-powered competitive analysis showing how your AI visibility compares to competitors. Returns side-by-side scoring and identifies gaps in your GEO strategy.
  • generate_html_fix — Generate production-ready HTML fixes for issues identified in your audit (schema markup, meta tags, etc.). Returns copy-paste-ready code snippets.
  • generate_remediation_strategy — Get an AI-powered remediation plan with prioritised actions ranked by impact and effort.

BYOK Toolkit Tools

These tools require Pro or Enterprise plans and use your own API keys (Bring Your Own Key):

  • run_bulk_meta — Optimise meta titles and descriptions at scale using your OpenAI API key. Processes multiple pages in a single run and returns optimised meta tags.
  • run_bulk_schema — Generate and validate Schema.org JSON-LD markup across multiple pages. Detects page type automatically and generates appropriate schema.
  • run_localizer — Generate geo-targeted content for programmatic local SEO using your Gemini API key. Creates location-specific variations of your content.
  • run_counter_measure — Generate competitor counter-measure briefs (2-pass AI pipeline, ~2,000-word outputs). Analyses a competitor's content strategy and generates specific counter-tactics.

How It Works

InnotekSEO AI's MCP server uses the 2025-03 specification with Streamable HTTP transport at /api/mcp/sse.

Authentication

Two auth methods are supported:

  1. OAuth 2.1 PKCE — For interactive use in Claude.ai, the standard OAuth flow with PKCE handles authentication automatically. When you first use an InnotekSEO tool in Claude, you'll be prompted to authorise the connection. After that, the session persists until you explicitly revoke access.
  2. Enterprise API Keys — For programmatic access, JWT-verified API keys are available on the Enterprise plan. These keys are passed via the Authorization header and support scoped permissions so you can restrict which tools a given key can access.

Detailed Workflow Examples

Here are step-by-step workflows for common GEO tasks:

Full Site Audit and Remediation

This is the most common workflow — auditing a site and generating fixes in one conversation:

  1. Start by asking Claude: "Run a GEO audit on example.com and tell me what needs fixing."
  2. Claude calls get_geo_audit and returns your scores: Entity Clarity 4/10, Fact Density 12, Schema Completeness 25%.
  3. Review the results and ask: "Generate HTML fixes for the schema issues and missing Organisation markup."
  4. Claude calls generate_html_fix and returns production-ready JSON-LD you can add to your site's <head>.
  5. Ask for a prioritised plan: "Create a remediation strategy for improving my GEO Grade from D to B."
  6. Claude calls generate_remediation_strategy and returns a ranked list of actions with estimated impact.

Competitive Analysis and Counter-Measures

Use this workflow to understand your competitive position in AI search:

  1. Ask: "Research how my AI visibility on example.com compares to competitor1.com and competitor2.com."
  2. Claude calls get_competitor_research and shows a side-by-side comparison of Entity Clarity, Fact Density, and Schema Completeness scores.
  3. Follow up: "Generate a counter-measure brief for competitor1.com — they're outperforming us on Entity Clarity."
  4. Claude calls run_counter_measure and produces a detailed 2,000-word brief covering specific actions to close the gap.

Bulk Schema Deployment

For sites with many pages that need structured data:

  1. Ask: "Generate Schema.org JSON-LD for these 10 product pages" and provide the URLs.
  2. Claude calls run_bulk_schema, which automatically detects each page type (Product, Article, LocalBusiness, etc.) and generates validated JSON-LD for each one.
  3. Review the output and ask Claude to adjust any specific entries before deploying.

Local SEO Content Generation

For businesses targeting multiple geographic areas:

  1. Ask: "Generate localised content for our plumbing services page targeting Manchester, Leeds, and Birmingham."
  2. Claude calls run_localizer and generates geo-targeted variations with location-specific details, local landmarks, and area-relevant content.
  3. Each variation includes appropriate LocalBusiness schema with geo coordinates and area-served data.

Comparison with Traditional API Workflows

Before MCP, integrating GEO tools into your workflow meant:

  1. Logging into a dashboard, running an audit, and exporting results
  2. Copying audit data into a spreadsheet for analysis
  3. Manually writing schema markup based on the audit findings
  4. Switching back to the dashboard to validate the markup
  5. Creating a remediation plan in a separate document

With MCP, this entire flow happens in one conversation. You ask Claude to audit, it returns results inline, you ask for fixes, it generates them, you ask for a plan, it creates one — all without context-switching. The AI assistant maintains the full context of your audit throughout the conversation, so every follow-up question is informed by everything that came before.

The productivity gain is substantial. What previously took 2–3 hours of switching between tools can be completed in a 20-minute conversation.

Troubleshooting

Connection Issues

If Claude can't connect to the InnotekSEO MCP server:

  • Check your plan. Free users have access to core audit tools only. BYOK toolkit tools require Pro or Enterprise.
  • Re-authorise. OAuth tokens expire periodically. If you see an authentication error, try starting a new conversation and re-authorising when prompted.
  • Rate limits. Free plans are limited to a set number of audits per day. If you hit the limit, you'll receive a clear error message with your reset time.

Unexpected Results

  • Low Entity Clarity scores. This usually means AI models are struggling to identify your brand. See our Entity Clarity guide for the 7 dimensions that affect your score.
  • Schema validation errors. The run_bulk_schema tool validates JSON-LD before returning it. If you see validation errors, they typically relate to missing required fields. Ask Claude to fix the specific errors it identifies.
  • BYOK API key errors. Ensure your OpenAI or Gemini API key is valid and has sufficient credits. The tools will return clear error messages if the upstream API rejects the key.

Best Practices for MCP Conversations

  • Be specific about URLs. Always provide full URLs including https:// when asking for audits.
  • One site per conversation. For best results, keep each conversation focused on a single site or competitive analysis.
  • Ask for explanations. After receiving audit results, ask Claude to explain what each score means and why it matters. The AI has full context to provide detailed explanations.

Why MCP Matters for GEO

MCP integration means your GEO workflow lives inside your AI assistant. Instead of switching between tools, you can:

  • Audit a site, analyse results, and generate fixes in a single conversation
  • Research competitors and create counter-measure strategies in one flow
  • Bulk-optimise meta tags while discussing content strategy
  • Generate llms.txt files informed by audit data
  • Deploy Schema.org markup validated against your audit scores

This is built and maintained by Innotek Solutions Ltd, the team behind InnotekSEO AI.

Getting Started

The MCP integration is available on all InnotekSEO AI plans. Free users get access to basic audit tools, while Pro and Enterprise users unlock the full suite including BYOK toolkits.

To set up the integration:

  1. Visit innotekseoai.com and create an account (or use the free tier — no signup required for basic audits).
  2. In Claude.ai, connect to the InnotekSEO AI MCP server.
  3. Authorise the connection via OAuth when prompted.
  4. Start asking Claude to run GEO audits and generate fixes.

The entire setup takes under 2 minutes, and you can run your first audit immediately.