# my-websearch [Health: Active]

**Category:** 🔎 Search & Data Extraction  
**Repository:** https://github.com/wtznicy/my-websearch  
**GitHub Stars:** 2  
**npm Downloads (last month):** 4009  
**Views:** 7  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/my-websearch

## Description
MyWebSearch - multi-engine web search MCP server, CLI and local daemon. No API keys required.

## Tools
Capabilities this server exposes over MCP:

- **search** — Search the web across multiple engines (Chinese/English/Code). Supports multi-query fan-out ('queries'), timeRange filtering, site limits, fetchTop article content, and no API keys required.
- **fetchWebContent** — Fetch readable web content/docs/PDFs from public URLs. Auto-cleans ads, keeps code/tables. Batch urls (1-3), CSS selector scoping, bm25 focus, startIndex paging, cache, schema extract to JSON, citations, link retention.
- **fetchGithubContent** — Fetch GitHub/Gitee by URL: README, source file (/blob/, with lines), tree, issue, PR, commit, compare, or releases. Example: https://github.com/o/r/blob/main/src/index.ts — ref can be branch/tag/commit; not a repo clone.
- **queryDocs** — Get official docs, API references, and code snippets for libraries/packages (e.g. 'react', 'fastapi', 'next.js'). Queries Context7 with fallback to npm/PyPI registry and GitHub README. Direct, no proxy needed.

## Claude Desktop Quick Installation
No confirmed install details yet, so no config is published. Follow the setup steps in the repository above.

## Documentation

## What my-websearch does

my-websearch provides MCP tools for search and data extraction, with additional CLI and local HTTP daemon modes. It can query a broad set of general, regional, news, developer, academic, and code-search sources, then return results with resolved links and structured metadata. The server is designed for AI coding workflows that need both discovery and retrieval rather than search-result links alone.

Its four MCP tools are `search`, `fetchWebContent`, `fetchGithubContent`, and `queryDocs`. Core operation is keyless. Exa can be used when an optional free key is configured, and Brave offers an optional API-key fallback.

## How it works

`search` routes requests among Chinese, English, and specialist engines. It supports multiple queries in one call, URL deduplication, time-range filtering, site restrictions, and an option to fetch the leading articles. Available sources include Bing, Baidu, CSDN, Juejin, Sogou, DuckDuckGo, Startpage, Yahoo, Brave, Exa, GitHub, Stack Overflow, Hacker News, arXiv, Bing News, and grep.app through the `code` search path.

The server uses direct HTTP retrieval and engine-specific request handling. It can detect operating-system proxy settings and route selected overseas engines through a proxy while keeping domestic engines direct. Outside mainland China, direct access without a proxy is the supported default. Some anti-bot challenges are handled computationally; DuckDuckGo's human CAPTCHA is reported rather than bypassed.

## Setup and configuration

The server supports MCP over STDIO for desktop clients and Streamable HTTP or SSE for remote clients. It can also run as a one-shot CLI command or a persistent local HTTP daemon with connection-pool reuse and a Prometheus `/metrics` endpoint.

No API key is needed for the main tools. `PROXY_URL` can be used when a proxy is required. `BRAVE_API_KEY` is an optional fallback for Brave search, while Exa supports an optional free key. The server also detects system proxy configuration on Windows, macOS, and Linux.

## Tools and capabilities

- `fetchWebContent` retrieves readable material from public web pages, documentation, and PDFs. It removes common page chrome while retaining titles, code blocks, and tables.
- Web extraction supports batches of one to three URLs, CSS-selector scoping, BM25-focused output, paging with `startIndex`, caching, JSON extraction through schemas, citations, and retained links.
- `fetchGithubContent` reads GitHub and Gitee repositories, source files, trees, issues, pull requests, commits, comparisons, releases, and README files. Source requests can select a branch, tag, or commit and can restrict output to line ranges.
- `queryDocs` looks up official library and package documentation, API references, and code snippets through Context7, with fallback lookup through npm, PyPI, and GitHub README content.
- Search results can use specialist APIs such as GitHub, Stack Exchange, Hacker News, arXiv, and grep.app in addition to conventional web engines.

## Limitations and notes

`fetchGithubContent` retrieves selected repository resources; it is not a repository clone. Web extraction targets public URLs. Engine availability and response behavior can vary with network conditions and anti-bot systems. Brave may require waiting, changing the IP, or using its API key when both HTTP retrieval and the browser-based challenge path fail. Exa is skipped when its optional key is not configured.

_Full upstream README: https://allmcps.com/mcp/my-websearch/readme_