MCP servers who are worth knowing if you are building AI | WebFlow -Blog

They let you pass on structured, scoped context to a model: files, metadata, app status, user roles and more. If you build something that goes beyond a rough prompt, such as agents, scoped assistants or on LLM -based code tools, you will probably use one.

This list focuses on the MCP servers that I recommend for building and debugging workflows in code bases, design systems, documentation and structured content. Each is available for integration, with open source code or documented APIs. I use Cursor If my primary IDE, so the installation instructions, focus on it, although most of these MCPs also work in other environments (Visual Studio Code, Windsurf, etc …). Cursor makes it easy to integrate MCP servers directly from the box with built-in support for adding and managing them in settings.

Which MCP servers should know when building AI

Context7

Context7 is a lightweight, open-source MCP server optimized for code bases. This allows you to expose a scoped display of your project folder to each model and support access to hundreds of open-source libraries and dependencies. This makes it especially valuable for agents who have to refer to real packages, functions or examples in the most recent context.

Cursor also supports documentation files in institutions. But you often have to remind the AI in your prompt to refer to it. View on Github or Visit context7.com.

Tip: If you don’t want to add use context7 For each prompt you can define a simple rule, Cursor Settings > Rules Section in Cursor (or the equivalent in your MCP client) to Auto-Invoke Context7 over any code question:

[[calls]]
match = "when the user requests code examples, setup or configuration steps, or library/API documentation"
tool  = "context7"

Once the rule is set, context7 is automatically viewed on relevant instructions. You can adjust the match pattern to cover your use cases. To see Documents.

Context7 important functions

• Including access to hundreds of open-source libraries such as Respond” Tailwind” Next.jsAnd Scare For up-to-date reference and code concept
• Cli -Tools and hosted version available
• Helps with avoiding most code examples and hallucinated APIs by training an accurate library context

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "context7": {
      "url": "https://mcp.context7.com/mcp"
    }
  }
}

Figma MCP

With the MCP server from Figma, your LLM or AI agent can gain access to structured design context directly from the DEV mode access. The surface component hierarchies, design tokens, selected nodes, variants and file structure, making it ideal for prompt-to-code workflows, spec diffs and automated design transfer. It is still in beta and runs locally in the Figma -Desktop app.

To call in the MCP server in Figma:

1. Open the Figma Desktop app and update to the latest version.
2. Make or open a Figma design file.
3. Open the Figma menu in the top left corner.
4. Select under preferences Call in the DEV mode MCP server.

Once switched on, you can use compatible tools to gain access to structured context from the DEV mode. Read the full guide or See the announcement For more information.

Figma MCP CHRISTMAS CHARACTERISTICS

• Structured design access Light component trees, tokens definitions, variants and selected junctions from DEV mode.
• Local DEV environment runs in the Figma Desktop app; No external API or cloud service required.
• Prompt conscious agents supports selection-based and link-based prompts in tools such as Cursor.
• Natural language editing says “Make this responsive” or “Center this text”. The agent applies the changes directly.
• Layout-conscious refactoring Agents can therefore apply automatic layout, restructuring frames and distance distance – not only adjust individual elements.
• Design-to-code Sync so that agents can update code bases or refactor components based on the current Figma file.
• Session Memory follows selections, assignments and intention about steps, so that agents do not lose any context halfway through the task.

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "Figma": {
      "url": "http://127.0.0.1:3845/sse"
    }
  }
}

Notion MCP server

The MCP server from Notion gives models structured access to content in your workspace. You can use it to build agents who summarize, clean up or generate content based on filters or templates. See MCP Supported tools from Notion

Notion MCP Server Key functions

• Search your notion of workspace using structured filters
• Make new pages from search results directly
• Update properties on multiple pages
• Context is checked for your access authorizations

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "Notion": {
      "url": "https://mcp.notion.com/mcp"
    }
  }
}

FirecrawL MCP

Firecrawl is a web crawling tool that gives your LLM structured access to real-time page content. It picks up a site, breaks it in semantic chunks and gives back clean context that your agent can use to reason about layout, metadata and content. Handy for product research, competition analysis or building content -conscious agents. Explore Firecrawl or View the Github Repo.

FirecrawL MCP key functions

• Web scraping, crawling and discovering content
• Search, content extraction and semantic chunking
• Batch research and deep crawl support
• Can walk in the cloud or host yourself
• Supports Server-Events Events (SSE) for streaming results

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "firecrawl-mcp": {
      "command": "npx",
      "args": ["-y", "firecrawl-mcp"],
      "env": {
        "FIRECRAWL_API_KEY": "YOUR-API-KEY"
      }
    }
  }
}

Github MCP server

The Github MCP server gives models structured access to your local repo, including diffs, commit history, file trees and more. It is designed to feed assistants who assess Pull requests, inline comments and write reasons about real code changes. View the Repo on Github.

Github MCP Server Key functions

• Browse, ask and understand code in your repositories
• Create and manage problems, Trek requests and project signs
• Monitor GitHub Actions Workflows and Problems Solving Builds
• View security reports and analyze code patterns
• Follow discussions, team activity and reports
• Including official installation guide for Cursor setup

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "github": {
      "url": "https://api.githubcopilot.com/mcp/",
      "headers": {
        "Authorization": "Bearer YOUR_GITHUB_PAT"
      }
    }
  }
}

Browserbase MCP server

The Browserbase MCP server connects your LLM with a live browser session in the cloud. It flows the full page context directly from an external browser, including DOM structure, cookies, local storage and navigation history, so that agents can work with the actual status of a page. This is useful for tools that simulate browse, extract data or user flows.

The MCP streams of browserbase streaming from the browser, including stupid structure, cookies, local storage and navigation history, so that your agent can work with the actual status of a page. See the MCP Integration Guide

Browserbase MCP Server Key functions

• Live Dom and Browser environment
• Contains session data, cookies, URL path and scroll position
• Compatible with browsing without headly or full -render pipelines
• Built for AI agents, test bots or UX simulations

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "browserbase": {
      "url": "your-smithery-url.com"
    }
  }
}

Or for more advanced functions:

{
  "mcpServers": {
    "browserbase": {
      "command": "npx",
      "args": ["@browserbasehq/mcp-server-browserbase"],
      "env": {
        "BROWSERBASE_API_KEY": "",
        "BROWSERBASE_PROJECT_ID": "",
        "GEMINI_API_KEY": ""
      }
    }
  }
}

Pandadoc MCP server

You may not use this every day, but it is one of the most practical ones. With Pandadoc you can build, send, send and manage digital documents such as supplier agreements, NDAs, invoices, proposals and contracts. The MCP server reveals the structured context of templates, fields and recipient flows so that models can help in generating, validating or routing documents. Read the Pandadoc -Documentation

Pandadoc MCP Server Key functions

• Access to document templates, field reports and roles
• Real-time preview of document state
• Good suitable for contract workflows, citation tools or approval systems
• Built on top of the public API of Pandadoc

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "pandadoc": {
      "url": "https://developers.pandadoc.com/mcp"
    }
  }
}

WebFlow MCP server

Web flow is a website experience platform (WXP) with which teams of marketers, designers and developers can visually build, manage and optimize websites. It brings design, data and implementation into a single environment that is built for production.

The WebFlow MCP server is used on the platform to provide electricity, CMS generation and AI-assisted-site updates. It defines structured context for the surface in which you work, including the designer and CMS. Explore WebFlow’s AI tools.

WebFlow MCP Server Key functions

• Access to CMS collections, items and Metadata page
• Make a list, pick up and publishing sites and pages
• Change styles, elements and scripts by structured tools
• Contains token-based auth, scoped tool access and model memory configuration
• Built for prompt-to-structure workflows such as scaffolding pages or updating SEO fields

Adaptation ~/.cursor/mcp.json Direct:

{
  "mcpServers": {
    "webflow": {
      "command": "npx",
      "args": ["-y", "webflow-mcp-server"],
      "env": {
        "WEBFLOW_TOKEN": "YOUR_API_TOKEN"
      }
    }
  }
}

How to add MCP servers to Cursor

Cursor supports native integration with MCP servers. You can manage it via the built-in MCP tool menu. Here are the steps to add MCP servers to the cursor.

1. Open cursor
2. Go to Settings → Tools and Integrations → MCP -Tools
3. Click Add new MCP server
4. Stick the MCP -Eindpunt -Eurl
5. Save and test the connection

All servers mentioned above work out of the box. No plug -ins or extra configuration required. You can also See the full cursor MCP -Setup guide.

Last thoughts

MCPs are not only infrastructure for AI. They give your LLM structure, reach and context so that it can take meaningful action between tools. Whether you pass on the onion status, CMS models, file diff or stupid data, these servers let you go beyond fast experiments to actual workflows. View this collection of MCP servers For other examples that are not here together with Cursor’s MCP -Directory.

If you want to look at what this looks like in a real DEV environment, here is a discount that runs through how agents use the Webflow MCP server to update SEO tags, locate CMS content and highlight new collections with just one prompt.