MCP servers connect AI assistants like Claude, Cursor, and ChatGPT to live tools and data via the Model Context Protocol. Browse vetted MCP servers and registries for web scraping, browser automation, search, databases, memory, and dev workflows.
Our editors' top MCP Servers picks for 2026 are Firecrawl (best for Turning any website into clean, LLM-ready data for your agents), Notion AI (best for Letting agents read and update pages in your Notion workspace), and Linear (best for Letting agents create, update, and triage Linear issues). All 56 tools in this category are hand-reviewed and re-checked each edition — the full ranked directory is below.








Workflow automation for everyone. Connects your tools to streamline mission-critical processes

The open-source ML hub — 2M+ models, 500k+ datasets, Spaces, Inference Endpoints, and the Transformers library.

Payments infrastructure for the internet — with an official MCP server and agent toolkit so AI agents can work with the Stripe API in natural language.

Explore advanced text-to-speech and voice cloning software for lifelike voiceovers and content generation.

Redis is an in-memory data store used as a vector database, semantic cache and memory layer for AI and agent applications.

The platform where the world builds software — repos, pull requests, Actions CI/CD, and Copilot — with an official MCP server so AI agents can work in your repositories.

Workflow automation platform for technical users.

A low-code platform for building next-gen apps and operationalizing data.

Neo4j is a graph database that powers knowledge graphs and GraphRAG so AI apps can ground answers in connected, verifiable relationships.

AI-powered writing assistant inside Notion to work faster, write better, and think bigger.

Unified platform for data analytics and AI innovation

AI work platform with native agents — Agent Factory builds custom AI agents, monday vibe turns prompts into apps. The work OS at $10B+ valuation.

Global edge network for deploying, securing, and speeding up apps — CDN, Workers, R2, and Workers AI — with an official MCP server for AI agents.

Online graphic design platform for creating various visual content.

Asana is a work management tool for planning, tracking, and automating projects — now with Asana AI for status summaries and AI teammates.

Cloud headless browsers for AI agents — production-grade infrastructure for web automation, scraping, and agent workflows.

Cloud platform for running, deploying, and scaling machine learning models with ease.

Create professional, custom websites with no code.
Jira is an agile project management tool from Atlassian for software teams — with Atlassian Intelligence and Rovo AI agents for issue work.

Developer platform to build, test, and deploy advanced voice agents in minutes.

Open-source vector database for storing data objects and vector embeddings

#1 iPaaS for enterprise AI and automation — Enterprise MCP for agentic AI, low-code integrations across hundreds of apps, secure and scalable.

Serverless Postgres with instant branching and pgvector — with an official MCP server so AI agents can create databases and run queries in natural language.

Build internal tools fast with Retool's platform.

Visual AI automation platform — build workflows and AI agents across 2,000+ apps.

Open-source vector database and search engine.

Pinecone: Transforming Vector Search for Enhanced Data Retrieval

Error monitoring and performance tracing for apps — with an official MCP server so AI agents can pull and analyze your Sentry issues to help debug.

AI search API and engine that retrieves the best, real-time web data for AI apps.

Full-stack web scraping and automation platform.

Grafana Labs is a full-stack observability platform with an AI assistant for dashboards and incident resolution.

MongoDB Atlas Vector Search adds semantic vector search to your database for RAG and AI agents.

Open-source AI browser automation SDK from Browserbase — write resilient browser agents using natural language with act, extract, observe, and agent primitives.

Real-time web search, extract, and crawl APIs built specifically for AI agents and RAG.

Auth and tool integrations for AI agents — connect any agent to 200+ apps with one SDK.

Streamlined tool for managing issues, sprints, and product roadmaps.

Product analytics, session replay, feature flags, and experiments — with an official MCP server so AI agents can query your product data in natural language.

Public MCP server to query the AI Tool Directory catalog from your AI agent — search, compare, find alternatives, and check whether a tool is still alive.

Honeycomb offers observability solutions to debug software in distributed environments.

Open-source secure sandboxes for AI-generated code execution — used by Claude, Perplexity, Hugging Face.

Type-safe ORM for Node.js and TypeScript, plus Prisma Postgres — with an official MCP server so AI agents can manage databases in natural language.

Open-source memory layer for AI agents — persistent personalization across sessions.

Browser API built for AI agents — anti-bot, sessions, and proxies designed for autonomous web browsing.

Build and deploy powerful applications connecting APIs, AI, databases, and more.

Web platform for modern frontend teams. Edge-deployed hosting, serverless functions, and AI-assisted development. 5M+ developers; $200M+ raised.

Context7 is an MCP server by Upstash that feeds AI code editors current, version-specific library docs and examples to cut hallucinated APIs.

Bright Data is a web data platform that provides proxies, scraper APIs, datasets, and a Web MCP server so developers and AI agents can collect public web d

Open-source dev environment manager that spins up containerized workspaces for code agents.

Figma prompt-to-app builder that turns ideas into working interactive prototypes.

Browserless is headless-browser infrastructure you can self-host via Docker or run managed — Puppeteer, Playwright, scrape, screenshot, and PDF APIs.

Web scraping built for LLMs — turn any website into clean markdown or structured data with one API call.

AI-native open-source embedding database for efficient data handling.

Zep is a memory platform for AI agents that builds temporal knowledge graphs from chat and business data so agents recall context that changes over time.

Hyperbrowser is cloud headless-browser infrastructure for AI agents, with stealth fingerprinting, proxy rotation, and built-in CAPTCHA solving.

AI-native CRM built for the next generation of B2B teams — embedded AI, flexible data modeling, deep workflows. The modern CRM picked by AI-native startups.

Arcade is the tool-calling platform for AI agents with built-in auth, observability, and DX. Founded by ex-Okta engineers; agent infra.
An MCP server is a small program that exposes one tool or data source — a database, a website scraper, a project tracker, a payments API — to an AI assistant through the Model Context Protocol (MCP). The assistant (the MCP client) connects to the server and can read data or take actions through it, without a custom integration for each app.
The Model Context Protocol is an open standard, introduced by Anthropic in late 2024, for connecting AI assistants to external tools and data. It defines a common way for an AI client such as Claude, Cursor, or ChatGPT to discover and call capabilities a tool exposes, so the same server works across any MCP-compatible app.
Many products now ship an official MCP server, grouped by the job they do: web data (Firecrawl, Bright Data, Apify), browser automation (Browserbase, Browserless, Steel), search (Exa, Tavily), databases (MongoDB, Redis), project tools (Notion, Linear, Asana, Monday, Airtable), design (Figma, Canva), and agent memory (Mem0, Zep). The right one depends on what you want your agent to do.
Most tools include their MCP server with an existing plan and many offer a free tier; usage of the underlying product is billed as normal. A few hosted servers meter calls separately. Check each provider's pricing page for current limits.
Most MCP clients — Claude Desktop, Cursor, Cline, and others — let you add a server in their settings or a JSON config file by pointing to the server's command or remote URL. Hosted providers give you a ready-made endpoint and key. Once added, the client lists the server's tools and your assistant can call them in a conversation.
An MCP server provides capabilities — tools, data, or actions — while an MCP client is the AI app that consumes them. Claude, Cursor, and most coding assistants are clients; the tools they connect to (Notion, GitHub, a database) are the servers. A single client can connect to many servers at once.
Receive weekly updates so you can stay up-to-date with the world of AI
Receive weekly updates so you can stay up-to-date with the world of AI