# IntellDirectories' MCP Server: Claude as a Directory Client At IntellDirectories, we've been clear about our mission: to make businesses globally AI-visible. This isn't just about indexing; it's about building the foundational infrastructure for a future where AI agents, not just humans, are primary consumers of business information.
Our latest step in this direction is the public release of our Model Context Protocol (MCP) server, specifically designed to empower large language models like Claude Desktop with real-time, structured directory data.
For too long, the promise of conversational AI has been hampered by a fundamental disconnect: LLMs are incredible at language, but notoriously poor at factual recall, especially when it comes to dynamic, real-world data.
They hallucinate, their knowledge cut-offs leave them perpetually out of date, and they lack direct access to the specific, verified details that businesses and consumers rely on daily.
We built the IntellDirectories MCP server to bridge this gap, transforming Claude from a generalist chatbot into a precise, tool-driven directory client. ## Bridging the Data Gap: Our 25 Tools The core of our MCP server, accessible at `/api/mcp`, is a robust suite of 25 distinct tools.
These aren't abstract concepts; they are concrete, callable functions designed to interact directly with our comprehensive business directory. Think of them as Claude's new senses, allowing it to perceive and interact with the real business world.
These tools cover a broad spectrum of directory functionalities, including: * `search_listings(query, location, category, limit)`: Find businesses based on keywords, location, and industry. * `get_aeo_score(business_id)`: Retrieve a business's AI Engagement Optimization score, indicating its visibility to AI agents. * `query_ai_visibility(business_id)`: Understand how a business ranks in AI-driven searches. * `fetch_reviews(business_id, sentiment, year, limit)`: Access customer reviews, filtered by sentiment or year. * `get_contact_info(business_id)`: Obtain phone numbers, email addresses, and websites. * `check_availability(business_id, date, time)`: Query opening hours or service availability. * `list_services(business_id)`: Discover specific services offered by a business. * `get_location_details(business_id)`: Fetch precise address and geographical data.
This comprehensive toolkit ensures that when Claude needs to answer a business-related query, it doesn't invent data; it retrieves it directly from the source.
We've engineered these tools for precision, speed, and reliability, providing Claude with the factual bedrock it needs to deliver accurate, actionable information. ## Connecting Claude Desktop in 30 Seconds We designed the integration process to be as straightforward as possible.
If you're a developer or just an enthusiastic Claude user, adding the IntellDirectories MCP server to your Claude Desktop instance takes less than a minute. Here’s how: 1. Open Claude Desktop. 2. Navigate to `Settings`. 3. Locate the `Model Context Protocol` section. 4. Click `Add Server`. 5. Paste `/api/mcp` into the URL field. 6. Confirm the addition. That's it.
Within seconds, all 25 of our specialized directory tools become available to Claude. This seamless integration is a testament to the power of open protocols and our commitment to making real-world data accessible. It's a common misconception that large language models are omniscient. They are, in fact, sophisticated pattern matchers and language generators.
Their true utility for real-world, current data *depends entirely* on robust external tools. Our MCP server turns Claude from a brilliant conversationalist into an effective, data-driven agent, proving that the future of AI isn't about isolated intelligence, but about intelligent orchestration of specialized services.
This isn't just an integration; it's a fundamental shift in how LLMs acquire and present factual information. ## Real-World Use Cases: Beyond Simple Search With the IntellDirectories MCP server connected, Claude's capabilities expand dramatically. The example "find me 5 vegan restaurants in Berlin with positive 2026 reviews" perfectly illustrates this.
Here's how Claude would compose an answer using our tools: 1. **Parsing the Request**: Claude identifies the core entities: "vegan restaurants," "Berlin," "5," and "positive 2026 reviews." 2. **Tool Selection**: It recognizes that `search_listings` is needed for the initial discovery and `fetch_reviews` for filtering. 3. **Executing `search_listings`**: Claude calls `search_listings(query="vegan restaurant", location="Berlin", limit=10)` (or a higher number to ensure enough results after filtering). 4. **Iterating and Filtering with `fetch_reviews`**: For each restaurant returned by `search_listings`, Claude would call `fetch_reviews(business_id=restaurant.id, sentiment="positive", year=2026)`.
If the `fetch_reviews` tool returns no reviews for 2026 for a given business (which is likely, given it's a future year), Claude would report this limitation or adapt, perhaps by asking if "recent positive reviews" would suffice, or by explaining that no businesses currently have reviews *from* 2026.
This demonstrates the tool's precision and Claude's ability to handle edge cases or clarify user intent based on factual limitations.
If the query meant "positive reviews *available in* 2026," Claude would simply use the `fetch_reviews(business_id=restaurant.id, sentiment="positive")` tool without a year filter, as all reviews would be "available in 2026." 5. **Composing the Answer**: Once 5 suitable restaurants are found (or the limitations are reported), Claude synthesizes the information into a clear, concise response, citing the data retrieved by the tools.
Consider other complex queries: * "What's the current AI Engagement Optimization score for 'The Green Bean Cafe' in London, and what are its top 3 positive reviews from the last 6 months?" This would involve `get_aeo_score` and `fetch_reviews` with specific parameters. * "Are there any businesses specializing in sustainable packaging solutions in San Francisco with an AI visibility score above 8.0, and what are their contact details?" This combines `search_listings`, `query_ai_visibility`, and `get_contact_info`. * "List the top 3 Italian restaurants in Rome with availability for a table for two next Saturday at 7 PM, based on customer ratings." This would orchestrate `search_listings`, `fetch_reviews` (for ratings), and `check_availability`.
These examples highlight how Claude, powered by our MCP server, moves beyond keyword matching to dynamic, intelligent data retrieval and synthesis, providing answers that are both accurate and contextually rich. ## The Future of Directory Interaction This isn't merely a technical integration; it represents a fundamental shift in how we conceive of directories and business information.
We are moving from a passive search model, where users manually sift through results, to an active, conversational paradigm where an AI agent acts as an intelligent intermediary. For businesses, this means the landscape of visibility is changing. Being listed in a traditional directory is no longer enough.
Your business needs to be structured, categorized, and optimized in a way that AI agents can easily parse and present. This is precisely what IntellDirectories provides: a platform engineered for AI visibility.
Our AEO scores and AI visibility metrics are not just arbitrary numbers; they are direct indicators of how effectively your business can be discovered and understood by the burgeoning ecosystem of AI clients. We believe that the best directories don't just list businesses; they empower them to connect with their audience in the most efficient ways possible.
By making our MCP server publicly available, we are accelerating this future, ensuring that your business is not just present online, but truly *AI-visible*. This is just the beginning. To ensure your business is part of this future, and visible to the next generation of AI clients, list with us today. [Add the connector →](/list-business)