Best for
- AI applications that need to fetch structured content from the web.
- Tools that need to process article or blog content.
- Applications that need YouTube transcripts.
- Scenarios that require proxy support and SSRF protection.
F
Fetch MCP Server
Community
A flexible HTTP fetching toolset with multiple formats and YouTube transcript support
GitHub source repository ↗
39FMRS · D
Fetch MCP Server provides a simple and powerful way for AI assistants to fetch and process web content. It supports multiple output formats and YouTube transcripts, offers proxy and SSRF protection, and is suitable for web data scraping and content extraction use cases.
Fetch MCP Server is a Model Context Protocol server for fetching web content in multiple formats, including HTML, Markdown, plain text, JSON, readable article content, and YouTube transcripts. It offers six tools that allow AI models to directly fetch and process web content. The server supports custom headers, proxies, pagination, and response size limits, and includes SSRF protection to block private/localhost addresses and DNS rebinding attacks.
Tools
fetch_html
Fetch a URL and return its raw HTML content.
fetch_markdown
Fetch a URL and return its content converted to Markdown.
fetch_txt
Fetch a URL and return plain text with HTML tags, scripts, and styles removed.
fetch_json
Fetch a URL and return the JSON response.
fetch_readable
Fetch a URL and extract the main article content using Mozilla Readability, returned as Markdown.
fetch_youtube_transcript
Fetch a YouTube video's captions or transcript.
Setup
- Ensure Node.js (or Bun) is installed.
- Add the following configuration to your MCP client:
{
"mcpServers": {
"fetch": {
"command": "npx",
"args": ["mcp-fetch-server"]
}
}
}- You can also use it as a CLI: install globally (
npm install -g mcp-fetch-server) and runmcp-fetch.
claude_desktop_config.json
{
"mcpServers": {
"fetch": {
"command": "npx",
"args": ["mcp-fetch-server"]
}
}
}
Fit and risk
Not for
- Not suitable for crawling that requires complex browser interactions (e.g., clicking, form filling).
- Not suitable for authenticated scraping that is not supported via custom headers.
- Not suitable for real-time streaming or long-running fetches.
Required permissions
- This server requires network access to fetch URLs.
- May need file system access for temporary storage or configuration.
- Ensure compliance with target websites' terms and policies.
Risks and side effects
- Potential violation of target website's terms or laws; use responsibly.
- Fetched content may contain malicious code or harmful content, but the server has SSRF protection to block private addresses.
- Potential performance issues due to large responses, but size limits are in place.
Troubleshooting
- If a request fails, check the URL and network connectivity.
- If using a proxy, ensure the proxy configuration is correct (e.g., `http://proxy:8080`).
- If YouTube transcripts fail, ensure yt-dlp is installed or check if the `lang` parameter is supported.
- If memory errors occur, adjust `max_length` or the `MAX_RESPONSE_BYTES` environment variable.
Use cases
Fetch web content directly within AI conversations for analysis or summarization.
Extract article body content, ignoring navigation and ads, for content summarization.
Retrieve YouTube video transcripts for translation, summarization, or search.
Fetch data from JSON APIs for data analysis and processing.
Convert web pages to Markdown or plain text for text processing.
Supported clients
Claude DesktopFull support
Claude CodeFull support
ClineFull support
ContinueFull support
GlamaFull support
Roo CodeFull support
WindsurfFull support