MCP-серверы
Открытый реестр серверов Model Context Protocol. Находите инструменты, ресурсы и промпты для ИИ-агентов по категории, типу подключения и сценарию использования.
Серверы
177
Инструменты
471
Категории
11
Участники
155
Official PayPal MCP server for commerce and payments. Lets AI agents create and capture orders, issue invoices, process refunds, manage subscriptions, and query transactions and disputes via the PayPal APIs. Useful for building checkout integrations, automating billing, and reconciling payments from an AI assistant.
Official Plaid MCP server for financial data connectivity. Lets AI agents work with Plaid's APIs to retrieve accounts, balances, and transactions, and to build and debug integrations for banking, payments, and identity. Useful for fintech development, personal finance tooling, and automating account data workflows.
MCP server for the Railway deployment platform. Lets AI agents create projects and services, deploy from repositories, manage environment variables, view deployment logs, and inspect service status. Useful for provisioning backends, databases, and cron jobs and for debugging deploys directly from an AI assistant.
MCP server for the Reddit API. Lets AI agents search subreddits, fetch hot/new/top posts, read comment threads, and retrieve user activity. Useful for market and community research, sentiment monitoring, trend discovery, and summarizing discussions across communities.
MCP server for Replicate, which hosts thousands of open machine learning models behind a single API. Lets AI agents search models, run predictions (image, video, audio, and text generation), poll prediction status, and retrieve outputs. Useful for adding generative media and specialized ML capabilities to agent workflows without managing infrastructure.
Official Semgrep MCP server for static application security testing. Lets AI agents scan code for security vulnerabilities and bugs, run custom rules, and return findings with severity and remediation guidance, embedding SAST into AI-powered development workflows.
MCP server for the Telegram Bot API. Lets AI agents send messages, deliver files and photos, and read updates from chats and channels through a bot, enabling notifications and conversational workflows from AI-powered tools.
MCP server for the Weaviate open-source vector database. Enables AI agents to store objects, run semantic and hybrid searches, and manage collections, making it a memory and retrieval backend for RAG applications directly from AI-powered tools.
MCP server that provides access to Wikipedia content. Lets AI agents search articles, fetch full or summarized page content, and resolve references, giving models reliable, citable background knowledge for research and question answering.
MCP server for the Wolfram Alpha computational knowledge engine. Gives AI agents access to step-by-step math, symbolic computation, unit conversions, scientific data, and curated facts across physics, chemistry, finance, and more. Useful for grounding quantitative answers and offloading precise computation from the model.
MCP server for Zendesk Support. Lets AI agents search and read tickets, add comments and replies, update ticket status and fields, and query help-center articles. Useful for building support copilots, triaging and summarizing tickets, and drafting grounded responses from an AI assistant.
Official Apify MCP server that gives AI agents access to thousands of Actors for web scraping and automation. Agents can run Actors to extract data from websites, search engines, social platforms, and maps, then retrieve structured results from the dataset, enabling rich web research and data collection.
MCP server for Atlassian Bitbucket that connects AI tools to repositories, pull requests, branches, and pipelines. Enables agents to review and create pull requests, read file contents and diffs, leave comments, and inspect build status on Bitbucket Cloud and Server.
MCP server for Atlassian Confluence that lets AI agents search the knowledge base, read and create pages, update content, and navigate spaces. Useful for grounding answers in internal documentation and for drafting or maintaining wiki content directly from an AI client.
MCP server for Gmail that lets AI agents read, search, draft, and send email, manage labels, and organize the inbox via the Gmail API with OAuth. Handy for triaging mail, drafting replies, and automating routine inbox workflows from an AI client.
MCP server for Google Calendar that lets AI agents view, create, update, and delete events, check availability across calendars, and schedule meetings via the Calendar API with OAuth. Useful for natural-language scheduling and calendar management.
Official Heroku MCP server that lets AI agents manage Heroku Platform resources. Supports listing and inspecting apps, scaling dynos, viewing logs, managing config vars and add-ons, and running one-off commands, so deployment and operations tasks can be handled conversationally.
MCP server for MySQL and MariaDB that lets AI agents inspect schemas, list tables, and run parameterized SQL queries. Supports read-only or read-write modes so agents can explore data, debug, and prototype against a MySQL database with controlled access.
Official SonarQube MCP server that brings code quality and security analysis into AI workflows. Lets agents fetch project issues, security hotspots, quality-gate status, and metrics from SonarQube Server or SonarCloud, so code health can be inspected and triaged conversationally.
MCP server for Todoist that lets AI agents create, update, complete, and query tasks using natural language. Supports projects, due dates, priorities, labels, and filters, making it easy to manage a personal or team task list conversationally.
MCP server for Trello that lets AI agents manage boards, lists, and cards. Supports creating and moving cards, updating due dates and labels, adding comments and checklists, and searching across boards, turning Trello into a conversational task and project tracker.
Skills и MCP-серверы
в чём разница?Skillsописывают, что делать
Skill объединяет инструкции, пример промпта и рекомендуемые модели, чтобы агент стабильно выполнял задачу. Skills добавляют знания, а не новые подключения.
MCP-серверыописывают, как подключиться
MCP-сервер расширяет возможности агента, подключая его к реальным системам: базам данных, API и файлам. MCP добавляет подключения и действия, а не инструкции к задаче.
Простое правило: выбирайте Skill, когда модели нужно хорошо выполнить задачу, и MCP-сервер, когда ей нужен доступ к инструменту или системе. Их можно сочетать: Skill может использовать инструменты MCP-сервера.
Создали MCP-сервер?
Добавьте его в открытый реестр, который поддерживает сообщество.