mcp-server-cloudflare-browser

An MCP server that fetches any URL and returns it as clean Markdown, powered by the Cloudflare Browser Rendering API. Useful for giving LLMs access to live web content without the noise of raw HTML.

Features

Eleven tools covering the full Cloudflare Browser Run Quick Actions API:

Tool	Description
`render_markdown`	URL → clean Markdown (JS-rendered)
`render_content`	URL → fully rendered HTML after JS execution
`take_screenshot`	URL → PNG screenshot
`render_pdf`	URL or HTML → PDF (base64)
`take_snapshot`	URL → multiple formats in one call (HTML + screenshot + markdown + a11y tree)
`get_accessibility_tree`	URL → accessibility tree (roles, names, states)
`extract_links`	URL → all links on the page
`scrape_elements`	URL + CSS selectors → structured element data
`extract_json`	URL + prompt/schema → AI-powered structured JSON
`start_crawl`	URL → async crawl job (follow links, up to N pages)
`get_crawl_status`	Poll crawl job status and retrieve results

Requirements

Node.js 18+
A Cloudflare account with Browser Rendering enabled
A Cloudflare API token with Workers AI or Browser Rendering permission

Quick Start

No installation needed. Run directly with npx:

CLOUDFLARE_ACCOUNT_ID=your-account-id \
CLOUDFLARE_API_TOKEN=your-api-token \
npx -y @vpsm-sh/mcp-server-cloudflare-browser

Configuration

Variable	Required	Description
`CLOUDFLARE_ACCOUNT_ID`	Yes	Your Cloudflare account ID (found in the dashboard sidebar)
`CLOUDFLARE_API_TOKEN`	Yes	API token with Browser Rendering read permission

To create an API token: Cloudflare Dashboard → My Profile → API Tokens → Create Token → use the Read all resources template or create a custom token with Browser Rendering access.

Usage with Claude Desktop

Add this to your claude_desktop_config.json:

{
  "mcpServers": {
    "cloudflare-browser": {
      "command": "npx",
      "args": ["-y", "@vpsm-sh/mcp-server-cloudflare-browser@latest"],
      "env": {
        "CLOUDFLARE_ACCOUNT_ID": "your-account-id",
        "CLOUDFLARE_API_TOKEN": "your-api-token"
      }
    }
  }
}

Config file location:

macOS: ~/Library/Application Support/Claude/claude_desktop_config.json
Windows: %APPDATA%\Claude\claude_desktop_config.json

Usage with VS Code

Add this to your .vscode/mcp.json (or user-level settings.json under "mcp.servers"):

{
  "servers": {
    "cloudflare-browser": {
      "command": "npx",
      "args": ["-y", "@vpsm-sh/mcp-server-cloudflare-browser@latest"],
      "env": {
        "CLOUDFLARE_ACCOUNT_ID": "${input:cloudflareAccountId}",
        "CLOUDFLARE_API_TOKEN": "${input:cloudflareApiToken}"
      }
    }
  }
}

VS Code will prompt for the values on first connect.

Usage with OpenCode

Add to your ~/.config/opencode/config.json:

{
  "mcp": {
    "cloudflare-browser": {
      "type": "local",
      "command": ["npx", "-y", "@vpsm-sh/mcp-server-cloudflare-browser@latest"],
      "env": {
        "CLOUDFLARE_ACCOUNT_ID": "your-account-id",
        "CLOUDFLARE_API_TOKEN": "your-api-token"
      }
    }
  }
}

Available Tools

`render_markdown`

Fetches a URL using a headless browser and returns the rendered page as Markdown.

Parameter	Type	Required	Description
`url`	`string`	One of	The URL to fetch and render
`html`	`string`	One of	Raw HTML to convert to Markdown
`gotoOptions`	`object`	No	Control page load behavior (`waitUntil`, `timeout`)
`rejectResourceTypes`	`string[]`	No	Block resource types (e.g., `["image", "stylesheet"]`)

Example: "Fetch https://developers.cloudflare.com/browser-rendering/ and summarize it."

`render_content`

Fetches a URL and returns the fully rendered HTML after JavaScript execution.

`take_screenshot`

Captures a screenshot of a fully rendered webpage. Returns a PNG image.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL to screenshot or HTML to render
`screenshotOptions`	`object`	No	`fullPage`, `omitBackground`, `type`, `quality`
`selector`	`string`	No	CSS selector to screenshot a specific element
`viewport`	`object`	No	`width`, `height`, `deviceScaleFactor`

`render_pdf`

Generates a PDF from a webpage or custom HTML. Returns base64-encoded PDF data.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML to convert to PDF
`pdfOptions`	`object`	No	`format`, `landscape`, `printBackground`, `displayHeaderFooter`, `headerTemplate`, `footerTemplate`, `margin`, `scale`

`take_snapshot`

Captures multiple formats in a single request: HTML, screenshot, Markdown, and/or accessibility tree.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML to snapshot
`formats`	`string[]`	No	At least two of: `content`, `screenshot`, `markdown`, `accessibilityTree`

`get_accessibility_tree`

Captures the accessibility tree of a webpage after JavaScript execution.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML
`interestingOnly`	`boolean`	No	Only semantically meaningful nodes (default: true)
`root`	`string`	No	CSS selector to anchor to a subtree

`extract_links`

Extracts all links from a webpage, including hidden ones.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML
`visibleLinksOnly`	`boolean`	No	Only return visible links (default: false)
`excludeExternalLinks`	`boolean`	No	Exclude external domain links (default: false)

`scrape_elements`

Extracts structured data from specific elements using CSS selectors.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML
`elements`	`object[]`	Yes	Array of `{ selector: string }`

`extract_json`

Extracts structured JSON data from a webpage using AI.

Parameter	Type	Required	Description
`url` or `html`	`string`	Yes	URL or HTML
`prompt`	`string`	No	Natural language prompt describing what to extract
`response_format`	`object`	No	JSON schema for structured output
`custom_ai`	`object[]`	No	Custom AI models with fallback support

`start_crawl`

Initiates an async crawl job that follows links across a site.

Parameter	Type	Required	Description
`url`	`string`	Yes	Starting URL
`limit`	`number`	No	Max pages (default: 10, max: 100000)
`depth`	`number`	No	Max link depth
`formats`	`string[]`	No	`html`, `markdown`, or `json` (default: html)
`render`	`boolean`	No	Use headless browser (default: true)
`includePatterns`	`string[]`	No	Wildcard patterns to include
`excludePatterns`	`string[]`	No	Wildcard patterns to exclude

`get_crawl_status`

Checks status or retrieves results of an async crawl job.

Parameter	Type	Required	Description
`jobId`	`string`	Yes	Crawl job ID from `start_crawl`
`status`	`string`	No	Filter records by status
`cursor`	`string`	No	Pagination cursor
`limit`	`number`	No	Max records (use 1 for status checks)

Development

# Install dependencies
npm install

# Compile TypeScript
npm run build

# Run the server
npm start

# Test interactively with the MCP Inspector
npm run inspector

After editing src/index.ts, always run npm run build before testing.

Testing with MCP Inspector

npm run inspector

This opens a web UI where you can call render_markdown manually and inspect requests/responses.

Adding Tools

See AGENTS.md for the full code style guide and tool-handler contract.

Name		Name	Last commit message	Last commit date
.github/workflows		.github/workflows
skills		skills
src		src
.gitignore		.gitignore
AGENTS.md		AGENTS.md
LICENSE		LICENSE
README.md		README.md
open-mcp.json		open-mcp.json
package.json		package.json
pnpm-lock.yaml		pnpm-lock.yaml
tsconfig.json		tsconfig.json

Navigation Menu

Repository navigation

About

Repository files navigation

mcp-server-cloudflare-browser

Features

Requirements

Quick Start

Configuration

Usage with Claude Desktop

Usage with VS Code

Usage with OpenCode

Available Tools

`render_markdown`

`render_content`

`take_screenshot`

`render_pdf`

`take_snapshot`

`get_accessibility_tree`

`extract_links`

`scrape_elements`

`extract_json`

`start_crawl`

`get_crawl_status`

Development

Testing with MCP Inspector

Adding Tools

License

About

Releases

Packages

Contributors

Languages

Latest commit

History

Folders and files

Repository files navigation

mcp-server-cloudflare-browser

Features

Requirements

Quick Start

Configuration

Usage with Claude Desktop

Usage with VS Code

Usage with OpenCode

Available Tools

render_markdown

render_content

take_screenshot

render_pdf

take_snapshot

get_accessibility_tree

extract_links

scrape_elements

extract_json

start_crawl

get_crawl_status

Development

Testing with MCP Inspector

Adding Tools

License

About

Resources

Stars

Watchers

Forks

Releases

Packages

Contributors

Languages

`render_markdown`

`render_content`

`take_screenshot`

`render_pdf`

`take_snapshot`

`get_accessibility_tree`

`extract_links`

`scrape_elements`

`extract_json`

`start_crawl`

`get_crawl_status`