MCPFind indexes 4,945 MCP servers organized into 21 categories. Each category represents a functional area: what the servers in it help an AI agent do. This article walks through every one of them with real numbers and named examples.
If you want to understand MCP before exploring categories, start with what is MCP. If you already know what category you need and want help picking a specific server, read how to choose the right MCP server. For a deeper look at what MCP servers actually are and how they work, MCP servers explained covers the concept in plain terms.
The Largest Categories: Developer Tools and AI/ML
Two categories dominate the directory by sheer volume.
Developer Tools leads with 2,349 servers, more than all other categories combined. This reflects where MCP adoption started: developers building AI coding assistants who needed agents to interact with GitHub, run terminal commands, manage Docker containers, and work with CI systems. If you write code and want to make your AI editor more capable, this category is your starting point.
AI and ML is second at 683 servers. These connect agents to external AI services: OpenAI APIs, Hugging Face models, vector databases like Pinecone and Weaviate, and local inference tools like Ollama. They are most useful when you want to chain AI models together or store and query embeddings inside a workflow.
MCP Server Categories for Data Access: Search, Databases, and Cloud
Three categories handle how agents get information from the world.
Search and Knowledge has 416 servers covering Brave Search, Tavily, Google, Wikipedia, and web scraping tools. This is the category that gives agents live information beyond their training data. For any research or writing task where freshness matters, a search server is one of the highest-value first installs.
Databases has 227 servers for PostgreSQL, MySQL, SQLite, MongoDB, Redis, and more. The most starred database server in the entire directory is Supabase at 2,556 stars. Direct database access is the most common advanced use case after developer tools.
Cloud Providers has 113 servers. Cloudflare's MCP server is the standout here, along with AWS, GCP, and Azure integrations. These servers let agents provision resources, query logs, and manage deployments through natural language.
Communication, Filesystems, and Security
These three mid-sized categories cover the infrastructure of how work happens.
Communication has 92 servers for Slack, Discord, Teams, email, and SMS. For any workflow where surfacing information in the places your team already looks is important, a communication server is a fast win.
Filesystems has 61 servers covering local disk, Google Drive, Dropbox, and S3. These are the building blocks for agents that work with documents. Nearly every document-heavy workflow needs a filesystem server at some point.
Security and Auth has 82 servers for Vault, Auth0, Okta, and secrets managers. These are used most by platform and DevOps teams who want agents that can audit access or rotate credentials without granting broad permissions.
Documentation, Automation, and Testing
These three categories each sit between 29 and 50 servers, covering structured development workflows.
Documentation has 50 servers for Confluence, Gitbook, ReadMe, and Mintlify. Teams use these to keep docs in sync with code or to let agents generate changelogs automatically.
Automation and Workflow has 30 servers for Zapier, Make, n8n, and Pipedream. If you already have automation workflows running, these servers let agents trigger them from a conversation rather than through a web UI.
Testing and Quality has 29 servers for Playwright, Cypress, Jest, and Vitest. The use case is agents that can run test suites, interpret failures, and iterate on broken tests without the developer switching context.
Analytics, Monitoring, and Social
Three more categories round out the operational and content sides of software.
Analytics and BI has 25 servers for Google Analytics, Mixpanel, Amplitude, BigQuery, and dbt. Monitoring and Observability has 24 servers for Datadog, Grafana, Sentry, and PagerDuty. Social Media also has 24 servers, covering Twitter/X, LinkedIn, Instagram, Reddit, and Bluesky.
The Smaller but Useful Categories
The final seven categories are smaller in server count but cover important use cases.
E-Commerce has 20 servers for Shopify, WooCommerce, and BigCommerce. Finance has 20 servers, with Stripe being the most widely used. Productivity has 8 servers including Notion, Google Calendar, and task managers. CRM has 7 servers for Salesforce and HubSpot. Maps and Location has 3 servers covering Google Maps, Mapbox, and OpenStreetMap.
The "other" category collects servers that do not fit cleanly into any single domain: browser automation, IoT integrations, and experimental tools that are still finding their category.
Together, these 21 categories map the current state of what AI agents can be connected to. The ecosystem is growing fast, with devtools and AI/ML leading and every other category adding servers steadily.
Adam Bush