← Home

Guide

Best MCP Servers for OpenClaw in 2026

A curated list of the best Model Context Protocol (MCP) servers you can use with OpenClaw. From web search and browser automation to databases and design tools — these MCP servers give your AI agent real-world capabilities.

What Is an MCP Server?

An MCP server (Model Context Protocol server) is a lightweight service that exposes tools and data to AI agents through a standardized protocol. Think of MCP servers as plugins that let your AI agent interact with external systems — databases, APIs, browsers, file systems, and more.

MCP was created by Anthropic as an open standard for connecting AI models to the real world. Instead of each AI platform building its own plugin system, MCP provides a universal protocol that works across Claude, OpenClaw, Cursor, and other AI tools.

OpenClaw has native MCP support through ClawHub, which hosts over 5,700 MCP skills. You can also connect raw MCP servers directly to your OpenClaw instance.

How MCP Works with OpenClaw

When you install an MCP server in OpenClaw, your agent gains access to the server's tools. For example, installing the Playwright MCP server gives your agent the ability to automate browsers. The flow is:

  1. You install an MCP server (via ClawHub skill or manual config)
  2. OpenClaw discovers the server's available tools automatically
  3. Your agent can now use those tools when relevant to your requests
  4. Results from the MCP server are returned to the agent for processing

Best MCP Servers by Category

Web Search & Browsing

  • Tavily MCP — AI-optimized web search returning clean, structured results. Free tier with 1,000 searches/month. The most popular search MCP server for AI agents. See our Tavily setup guide.
  • Playwright MCP — Full browser automation using Playwright. Navigate pages, click elements, fill forms, take screenshots, and extract data from JavaScript-rendered sites. Works headlessly on servers.
  • Chrome DevTools MCP — Connect directly to Chrome's DevTools Protocol for debugging, performance profiling, and page inspection. More low-level than Playwright but powerful for web development tasks.
  • Firecrawl MCP — Web scraping and crawling service. Extract structured data from websites, crawl entire sites, and convert pages to clean markdown.
  • SearXNG MCP — Self-hosted meta-search engine. Aggregates results from multiple search engines with privacy focus. Good for users who want to avoid commercial search APIs.

Development & Coding

  • GitHub MCP — Full GitHub API access. Create repos, manage issues, review pull requests, search code, and automate workflows. One of the most-used MCP servers in the ecosystem.
  • Supabase MCP — Interact with Supabase databases, auth, storage, and edge functions directly from your AI agent. Create tables, query data, manage users, and deploy serverless functions.
  • Code Runner MCP — Execute code in sandboxed environments. Supports Python, JavaScript, TypeScript, and more. Safe execution with resource limits.

Design & Creative

  • Figma MCP — Read and interact with Figma designs. Extract design tokens, inspect components, and export assets. Useful for design-to-code workflows.
  • Image Generation MCP — Generate images using DALL-E, Stable Diffusion, or other AI models. Text-to-image, image editing, and style transfer.

Automation & Workflows

  • n8n MCP — Connect to n8n workflow automation platform. Trigger workflows, read execution results, and build complex automations that combine multiple services.
  • HTTP Request MCP — Make arbitrary HTTP API calls. Connect to any REST API, webhook, or web service. The Swiss army knife of MCP servers.
  • Cron Scheduler MCP — Schedule recurring tasks. Set up periodic actions like daily reports, weekly summaries, or automated monitoring.

Knowledge & Productivity

  • Obsidian MCP — Read and write to Obsidian vaults. Search notes, create new entries, manage tags, and build a knowledge base that your AI agent can access.
  • Notion MCP — Manage Notion workspaces. Create pages, update databases, search content, and automate Notion workflows.
  • Perplexity MCP — Access Perplexity AI's research capabilities. Get well-sourced answers with citations for research tasks.

Data & Databases

  • PostgreSQL MCP — Query and manage PostgreSQL databases. Run SQL, inspect schemas, and analyze data directly from your agent.
  • SQLite MCP — Lightweight database operations. Perfect for local data storage and simple analytics.

MCP Server Comparison Table

MCP ServerCategoryAPI Key RequiredFree Tier
Tavily MCPWeb SearchYes1,000 searches/mo
Playwright MCPBrowserNoFully free
GitHub MCPDevelopmentYes (PAT)Yes
Supabase MCPDatabaseYesYes (500MB)
Figma MCPDesignYesRead-only free
n8n MCPAutomationSelf-hostedYes (self-host)
Obsidian MCPKnowledgeNoFully free
Chrome DevTools MCPBrowserNoFully free

How to Install MCP Servers in OpenClaw

Method 1: ClawHub (Recommended)

The easiest way to add MCP servers to OpenClaw is through ClawHub:

/skill install @anthropic/tavily-search
/skill install @anthropic/playwright
/skill install @community/github

ClawHub handles dependencies, configuration prompts, and updates automatically.

Method 2: Manual Configuration

For MCP servers not on ClawHub, add them directly to your openclaw.json:

{
  "plugins": {
    "entries": {
      "my-mcp-server": {
        "enabled": true,
        "type": "mcp",
        "command": "npx",
        "args": ["-y", "@example/mcp-server"]
      }
    }
  }
}

MCP Protocol Explained

The Model Context Protocol (MCP) is an open standard created by Anthropic for connecting AI models to external tools and data sources. Key concepts:

  • Server — A process that exposes tools, resources, and prompts to AI agents. Each MCP server specializes in a specific domain (web search, database access, file management, etc.).
  • Client — The AI application (OpenClaw, Claude, Cursor) that connects to MCP servers and uses their capabilities.
  • Tools — Functions that the MCP server exposes. For example, a web search server might expose a search tool and a fetch_page tool.
  • Resources — Data that the MCP server makes available. For example, a database server might expose tables as resources.
  • Transport — MCP supports stdio (local process) and SSE (remote HTTP) transports. OpenClaw supports both.

MCP Servers on OpenClaw Launch

On OpenClaw Launch, popular MCP servers are available as one-click skills from the Dashboard. No manual configuration, no dependency management, and automatic updates.

  • Pre-tested MCP server compatibility
  • Visual configuration for API keys and settings
  • Automatic updates when new MCP server versions are released
  • No openclaw.json editing needed

Frequently Asked Questions

What is an MCP server?

An MCP server (Model Context Protocol server) is a lightweight service that gives AI agents access to external tools and data. MCP is an open standard by Anthropic that works across OpenClaw, Claude, Cursor, and other AI tools. MCP servers let your agent search the web, access databases, automate browsers, manage files, and more.

What is MCP in AI?

MCP (Model Context Protocol) is an open standard for connecting AI models to external tools and data sources. It provides a universal protocol so AI agents can interact with web APIs, databases, file systems, and other services. Think of it as USB for AI — a standard connector that works across different AI platforms.

What is the MCP protocol?

The MCP protocol (Model Context Protocol) defines how AI applications communicate with external tool servers. It uses a client-server architecture where AI apps (clients) connect to MCP servers that expose tools, resources, and prompts. MCP supports local (stdio) and remote (SSE/HTTP) connections and is maintained as an open standard by Anthropic.

How many MCP servers can OpenClaw use?

OpenClaw can connect to multiple MCP servers simultaneously. Through ClawHub, you have access to over 5,700 MCP-based skills. You can also add custom MCP servers alongside ClawHub skills. There is no hard limit on the number of MCP servers.

What are the best MCP servers for Claude?

The best MCP servers for Claude and OpenClaw include Tavily (web search), Playwright (browser automation), GitHub (code management), Supabase (database), and Obsidian (knowledge management). The choice depends on your use case — see the full curated list above.

How do I find MCP servers on GitHub?

Search GitHub for “mcp server” or “model context protocol server” to find community-built MCP servers. Many are published as npm packages (for Node.js/TypeScript) or Python packages. The awesome-mcp-servers repository maintains a curated list of quality MCP servers.

Related Guides

MCP Servers, One Click

Install top MCP servers from the Dashboard. No config files needed.

Deploy with OpenClaw Launch