Web Scout MCP
MCP server with DuckDuckGo search and page text extraction: several URLs can be passed in one call
- 151 ★ on the repository
- Last commit
Extended access
Why this level
- The server visits arbitrary pages and pulls their text into context, which may contain instructions aimed at the model
- Queries are visible to DuckDuckGo and the visited sites
Install
- Works in
In your terminal, with SkillFoxx CLI
npx skillfoxx add mcp/web-scout-mcpDetects the agents on your machine, shows the access level and pins the version.
Other ways to install
Access and checks
How we reviewExtended accessThe tool runs code or makes network calls and only reads external data.
- The server visits arbitrary pages and pulls their text into context, which may contain instructions aimed at the model
- Queries are visible to DuckDuckGo and the visited sites
This is third-party code. Review the repository files before installing.
What it does
Web Scout MCP gives the agent two tools. `DuckDuckGoWebSearch` searches DuckDuckGo and returns a list of titles, URLs and snippets, with `maxResults` setting the count and ten as the default. `UrlContentExtractor` fetches a page and strips scripts, styles and navigation, leaving readable text. The extractor takes one URL or an array of URLs, and pages are then processed in parallel. The server throttles requests so the search engine does not block them, watches memory and returns clear errors. It runs on Node.js 18 and newer over stdio with `npx -y @pinkpixel/web-scout-mcp@latest`, and no keys are needed.
Who it is for. Developers who need simple search and page reading for an agent with no sign-up and no keys.
Good fit when
- The agent looks for an error across the web and reads three found pages in one call
- You need a quick topic overview: search returns ten links and the extractor collects their text
- The client is Cursor or Claude Desktop and setup comes down to one entry in `config.json`
- Search must work without keys and without signing up for search APIs
Not a fit when
- You need semantically ranked results with passages: the server returns full page text
- You need hundreds of requests a minute: the rate limiter will slow results down
- You need pages built on JavaScript: the server reads fetched HTML without running scripts
Example request
Find three recent articles about the ECONNRESET error in Node.js through DuckDuckGo and extract their textMCP
- Transport
- stdio
- Authentication
- not required
Limitations
The server talks to DuckDuckGo and sites directly, so the result depends on search availability and on whether sites accept automated requests. Pages that depend on JavaScript are read incompletely because scripts do not run. The rate limiter protects against blocking but slows bulk requests. The package version and install paths are described in the README and the license is Apache 2.0. The last commit in the repository dates from November 1, 2026. The `WEB_SCOUT_DISABLE_AUTOSTART=1` variable turns off autostart and is needed when embedding in your own code.
How to disable. Delete the `web-scout` entry from the client `config.json` or run `claude mcp remove web-scout`, and remove a global package with `npm uninstall -g @pinkpixel/web-scout-mcp`.
FAQ
Is an API key required?
No, search goes through DuckDuckGo without keys and setup comes down to adding the server to the client config.
Which tools does the server have?
Two: `DuckDuckGoWebSearch` for search and `UrlContentExtractor` for extracting the text of one page or several.
How do I pass several URLs?
Pass an array of strings in the `url` field, for example two article addresses, and pages are processed in parallel.
How do I install the server?
Through Smithery with `npx -y @smithery/cli install @pinkpixel-dev/web-scout-mcp --client claude` or globally with `npm install -g @pinkpixel/web-scout-mcp`.
Which Node.js version is needed?
Node.js 18 or newer. The server starts over stdio when the published entry point runs.
Discussion
No questions or reviews yet.
Related
MCP serversWeb search via DuckDuckGo and page reading for the agent with no API keys, with built-in rate limiting
A web search MCP server that gives the agent full page text at once: Stack Overflow answers, GitHub threads, arXiv papers
Local web search for agents: searches, reads pages and returns compact passages with sources, saving context tokens
A research MCP server: searches the web, SEC, courts and patents, reads whole pages and checks that cited sources really exist.
A local MCP server with no API keys: multi-engine web search, page fetching and document reading tuned for LLMs.
A unified search MCP server: one interface to Tavily, Brave, Kagi, Exa, Linkup, Firecrawl and GitHub search
README in short
A short description, a feature list, three install methods, startup and client config, a description of two tools with input and output examples, a development section and a list of docs.