Best MCP Servers (2026)

1

Playwright MCP

Open-source MCP server that lets LLMs drive real browsers via Playwright and accessibility snapshots.

4.8 (6)

· free

Full review

Playwright MCP is an open-source Model Context Protocol server that exposes Playwright's browser automation capabilities to large language models. Instead of relying on screenshots and vision models, it surfaces structured accessibility snapshots of web pages, giving agents a fast, deterministic view of the DOM that they can reason over and act on. It lets LLM-powered agents navigate sites, click elements, fill forms, extract data, and run end-to-end workflows across Chromium, Firefox, and WebKit. Because it speaks MCP, it plugs into any compatible client such as Claude Desktop, Cursor, or custom agent frameworks, making real-world browser tasks accessible to autonomous and assisted workflows.

• MCP server interface for LLM agents
• Structured accessibility tree snapshots
• Cross-browser support via Playwright
• Click, type, navigate, and form-filling actions
• Headless or headed browser modes
• Integration with Claude, Cursor, and custom clients

Pydantic AI is an open-source Python framework for building applications powered by large language models. Created by the team behind Pydantic, it brings the same focus on type safety, validation, and developer ergonomics to agent development, making LLM outputs predictable and easier to integrate into production code. The framework supports multiple model providers, structured responses validated through Pydantic models, tool calling, dependency injection, and streaming. It is designed to feel familiar to Python developers and works well alongside existing stacks like FastAPI, making it suitable for everything from quick prototypes to production-grade GenAI services.

• Typed agents with Pydantic-validated outputs
• Support for OpenAI, Anthropic, Gemini, and more
• Tool and function calling with dependency injection
• Streaming responses and async-first design
• Integration with FastAPI and observability tools
• Testing utilities for deterministic agent behavior

Cognee is an open-source AI memory platform designed for AI agents. It provides a persistent long-term memory across sessions by ingesting data in any format and building a self-hosted knowledge graph. Cognee combines vector embeddings, graph reasoning, and cognitive-science-grounded ontology generation, making documents searchable by meaning and connected by evolving relationships. This platform is suitable for developers and organizations looking to unify data from various sources, enable domain knowledge in agents, and create reliable and trustworthy agents. Cognee offers features such as unified ingestion, graph and vector search, local operation, ontology grounding, multimodal capabilities, learning from feedback, context management, and cross-agent knowledge sharing. It also provides agentic user/tenant isolation, traceability, and audit traits. The platform supports multiple clients, including Python, Rust, and TypeScript, and is available as plugins for OpenClaw and Claude Code.

• Knowledge graph based agent memory
• Semantic and structured data ingestion
• Python SDK for agent integration
• Pluggable LLM and storage providers
• Querying across past sessions and documents
• Self-hosted or managed deployment options

4

Inbox Zero

AI email assistant that organizes, drafts replies, and helps you reach inbox zero faster.

4.8 (4)

· free

Full review

Inbox Zero is an AI-powered email assistant designed to help users manage their inbox more efficiently. It organizes emails, drafts replies, and manages calendars, with the goal of helping users reach "inbox zero" faster. The tool is accessible via a web interface and can also be interacted with through Slack or Telegram for on-the-go management. Inbox Zero is an open-source alternative to similar tools like Fyxer, offering more customization options and enhanced security features. Key features include an AI personal assistant that learns the user's tone and style to pre-draft replies, AI-driven rules for handling emails based on plain English instructions, and tools for tracking emails that require a reply or are awaiting responses. It also offers bulk unsubscription and archiving capabilities, blocks cold emails, and provides email analytics. Additionally, Inbox Zero can generate meeting briefs by pulling context from both email and calendar events and automatically save email attachments to cloud storage services like Google Drive or OneDrive. The tool is built using a range of technologies including Next.js, Tailwind CSS, and Prisma, and is hosted on GitHub. Users can choose between a hosted version available at getinboxzero.com or self-hosting using a CLI setup that requires Docker and Node.js. The project is active, with a community that contributes to its development and feature requests can be made via GitHub issues or the project's Discord channel. Inbox Zero aims to reduce the time users spend in their inbox, allowing them to focus on more important tasks. While it offers a comprehensive set of features for email management, the extent of customization and the learning curve for setting up and fully leveraging the AI capabilities may vary. Overall, Inbox Zero presents itself as a flexible and secure solution for individuals looking to automate and streamline their email and calendar management tasks.

• AI-generated reply drafts
• Automatic email categorization and prioritization
• Bulk unsubscribe from newsletters
• Custom automation rules
• Inbox analytics and cleanup tools
• Smart reminders for unanswered emails

5

Screenpipe

Open-source 24/7 local screen and audio recording for building context-aware AI apps

4.8 (4)

· free

Full review

Screenpipe is an open-source platform that continuously captures screen activity and audio on your device, storing everything locally so developers can build AI applications grounded in real user context. By indexing what you see, hear, and do, it provides a rich personal data layer that apps and agents can query without sending information to the cloud. The project targets developers creating productivity tools, memory assistants, meeting summarizers, and personalized agents. It exposes APIs and a plugin system so custom pipelines can transform raw recordings into searchable text, transcripts, and structured events that feed downstream LLM workflows. Because all processing happens on the user's machine, Screenpipe emphasizes privacy and ownership of data while remaining extensible through community-built integrations.

• 24/7 screen and audio capture
• Local storage and on-device processing
• OCR and speech-to-text indexing
• Plugin and pipeline architecture
• APIs for querying captured context
• Cross-platform desktop support

6

AgentKit

TypeScript library for building and orchestrating AI agents with tools, memory, and multi-agent workflows.

4.5 (4)

· free

Full review

AgentKit is an open-source TypeScript framework designed for developers who want to build production-ready AI agents without reinventing core orchestration logic. It provides primitives for defining agents, attaching tools, managing state, and coordinating multi-agent workflows in a type-safe way. The library focuses on composability, letting you chain agents, route tasks between specialists, and integrate with existing model providers and APIs. It fits naturally into Node.js and serverless environments, making it suitable for backend services, internal automation, and customer-facing AI features. Because it is code-first and unopinionated about UI, AgentKit is best suited to engineering teams comfortable working in TypeScript who want fine-grained control over how their agents reason, call tools, and handle long-running tasks.

• Agent and tool abstractions
• Multi-agent routing and handoffs
• State and memory management
• Model-agnostic provider support
• Type-safe APIs for inputs and outputs
• Works in Node.js and serverless runtimes

7

onchain-mcp

Bringing the bankless onchain API to MCP

No reviews yet· free

Full review

The Bankless Onchain MCP Server is a framework for interacting with on-chain data via the Bankless API. It implements the Model Context Protocol (MCP) to allow AI models to access blockchain state and event data in a structured way. The server provides various data operations, including contract state reading, event logs fetching, and transaction history retrieval. It caters to developers and researchers who need to interact with blockchain data in a structured manner. This project is no longer receiving updates, and its maintenance status might affect its stability and feature availability.

• Contract operations (read contract state, get proxy, get ABI, get source)
• Event operations (get events, build event topic)
• Transaction operations (get transaction history, get transaction info)

8

markitdown

Python tool for converting files and office documents to Markdown.

No reviews yet· free

Full review

MarkItDown is a lightweight Python utility for converting various files to Markdown for use with LLMs and related text analysis pipelines. It is most comparable to textract, but with a focus on preserving important document structure and content as Markdown, including headings, lists, tables, links, etc. The output is often reasonably presentable and human-friendly, but it is meant to be consumed by text analysis tools, and may not be the best option for high-fidelity document conversions for human consumption. MarkItDown currently supports the conversion from PDF, PowerPoint, Word, Excel, Images (EXIF metadata and OCR), Audio (EXIF metadata and speech transcription), HTML, Text-based formats (CSV, JSON, XML), ZIP files, Youtube URLs, EPubs, and more. It is recommended to use a virtual environment to avoid dependency conflicts. With Python 3.10 or higher, you can install MarkItDown using pip: pip install 'markitdown[all]' or from the source with: git clone git@github.com:microsoft/markitdown.git, then pip install -e 'packages/markitdown[all]'. The usage of MarkItDown involves command-line invocation, either by specifying the output file, piping content, or using the narrowest convert_* function for specific use cases.

• Conversion of PDF, PowerPoint, Word, Excel
• Support for Images (EXIF metadata and OCR)
• Support for Audio (EXIF metadata and speech transcription)
• Support for HTML, Text-based formats (CSV, JSON, XML)
• Support for ZIP files, Youtube URLs, EPubs
• Optional dependencies for activating various file formats

9

mcp-clickhouse

mcp-clickhouse MCP server

No reviews yet· free

Full review

The mcp-clickhouse MCP server is an An MCP server for ClickHouse. It features ClickHouse Tools, including run_query to execute SQL queries on your ClickHouse cluster, list_databases to list all databases on your ClickHouse cluster, and list_tables to list tables in a database with pagination. Additionally, it includes chDB Tools, like run_chdb_select_query, to execute SQL queries using chDB's embedded ClickHouse engine. It also provides a Health Check Endpoint to check the server's health and a Security mechanism for authentication. The server can be set up for internal services, local development, or with OAuth / OIDC authentication providers through FastMCP.

• run_query to execute SQL queries on ClickHouse cluster
• list_databases to list all databases on ClickHouse cluster
• list_tables to list tables in a database with pagination
• run_chdb_select_query to execute SQL queries using chDB's embedded ClickHouse engine
• Health Check Endpoint for server monitoring
• Multiple authentication modes, including OAuth and OIDC through FastMCP

MCP Server for QA Sphere TMS is a client used for integrating Large Language Models (LLMs) with QA Sphere (QSP) to boost test script creation capabilities. After configuring the server (details available on GitHub), LLMs can interact with QA Sphere's automated test cases. By leveraging the MCP (Model Callback Protocol), it enables developers and testers to quickly create AI-based test cases, automate tasks, and run test suites integrated with QAS Sphere. The MCP-based solution is user-backed, allowing a wide range of QA tasks to be automated, including test-case discovery and execution. Additionally, you can reference large language models, automate tasks, and run test suites integrated with the QA Sphere Test Management System.