@trishchuk/fetch MCP

A fetch tool for AI agents with browser-compatible network profiles, built on the Model Context Protocol · npm package @trishchuk/mcp-fetch-server · Use cases →

npm ↗ GitHub ↗ @trishchuk/fetch on GitHub ↗ @trishchuk/fetch on npm ↗

A high-performance Model Context Protocol (MCP) server providing a fetch tool for AI agents — Claude Code, Claude Desktop, Cursor, Windsurf, Cline, Antigravity, and others.

Powered by @trishchuk/fetch — a native curl-impersonate-style HTTP client that sends requests with browser-compatible TLS (JA3/JA4, ClientHello) and HTTP/2 profiles. This page calls the server @trishchuk/fetch MCP; it is published on npm as @trishchuk/mcp-fetch-server.

Why this server

Many sites classify HTTP clients by network characteristics — TLS and HTTP/2 fingerprints — before serving content, so standard Node.js/Undici clients are often refused on public pages that a browser opens normally. Standard MCP fetching tools also tend to fail on large payloads or overflow LLM context windows.

@trishchuk/fetch MCP addresses both problems:

Requirements

Installation & Setup

Run with npx (no installation needed)

npx -y @trishchuk/mcp-fetch-server

Global or local installation

# Global
npm install -g @trishchuk/mcp-fetch-server

# Or clone & install locally
git clone https://github.com/x51xxx/mcp-fetch-server.git
cd mcp-fetch-server
npm install

Claude Code

# Using npx (recommended)
claude mcp add fetch -- npx -y @trishchuk/mcp-fetch-server

# Or using a local path
claude mcp add fetch -- node /path/to/mcp-fetch-server/src/index.js

Claude Desktop

Add to claude_desktop_config.json:

{
  "mcpServers": {
    "fetch": {
      "command": "npx",
      "args": ["-y", "@trishchuk/mcp-fetch-server"]
    }
  }
}

Cursor / Windsurf / Antigravity (.mcp.json)

Create or update .mcp.json in your workspace with the same mcpServers block shown above.

Tool Reference: fetch

ParameterTypeDefaultDescription
urlstringrequiredTarget absolute URL (e.g. https://example.com/api).
methodstring"GET"HTTP method (GET, POST, PUT, DELETE, PATCH, HEAD, etc.).
headersobject—Request headers as key-value pairs.
bodystring—Request body sent as a UTF-8 string (JSON, form-encoded, raw text).
impersonatestring"chrome_147"Browser fingerprint preset (e.g. "chrome_147", "safari_26", "random").
platformstring—Declared OS: "windows", "macos", "linux", "android", or "ios".
proxystring—Proxy URL: http://, https://, or socks5:// (supports user:pass@host:port).
sessionstring—Session ID for sharing client connections and cookie jars across multiple calls.
resolveobject—Custom DNS pinning (e.g. {"example.com": "1.2.3.4"}). SSRF-safe testing.
redirectstring"follow"Redirect mode: "follow", "manual", or "error".
httpVersionstring—Force protocol version: "http1" or "http2".
tlsMinVersionstring—Minimum TLS version: "1.0", "1.1", "1.2", "1.3".
tlsMaxVersionstring—Maximum TLS version: "1.0", "1.1", "1.2", "1.3".
timeoutMsnumber—Overall request timeout in milliseconds.
maxResponseBytesnumber2097152Body limit in bytes (max 2MB). Responses exceeding this are safely truncated.
encodingstring"auto"Body return format: "auto" (text for textual MIME types, base64 for binary), "text", or "base64".

Response Schemas

Successful HTTP exchange

Any completed HTTP transfer returns a standard JSON result (including 404, 500, or 3xx under redirect: "manual"):

{
  "status": 200,
  "statusText": "OK",
  "ok": true,
  "url": "https://example.com/data",
  "redirected": false,
  "headers": {
    "content-type": "application/json; charset=utf-8",
    "cache-control": "max-age=3600"
  },
  "bodyEncoding": "text",
  "body": "{\"message\": \"Hello world\"}",
  "truncated": false
}

Network / transport failure

If the network connection fails, times out, or the URL is invalid, the tool returns isError: true:

{
  "error": true,
  "code": "TIMEOUT",
  "message": "failed to read response body: request or response body error: operation timed out"
}

Usage Examples for Agents

Request a page with a browser profile

{
  "url": "https://example.com/docs/api",
  "impersonate": "chrome_147",
  "platform": "macos",
  "headers": {
    "Accept-Language": "en-US,en;q=0.9"
  }
}

Multi-step requests with a persistent session (cookie jar)

// Step 1: sign in to your own account to obtain a session cookie
{
  "url": "https://example.com/api/login",
  "method": "POST",
  "session": "agent-session-01",
  "headers": { "Content-Type": "application/json" },
  "body": "{\"user\":\"admin\",\"password\":\"secret\"}"
}

// Step 2: request a page that needs the session (cookies are preserved automatically)
{
  "url": "https://example.com/api/dashboard",
  "session": "agent-session-01"
}

Route through a SOCKS5 proxy

{
  "url": "https://example.com",
  "proxy": "socks5://user:[email protected]:1080",
  "impersonate": "safari_26"
}

Fetch binary assets (images, PDFs)

{
  "url": "https://example.com/report.pdf",
  "encoding": "base64"
}

DNS pinning for SSRF-safe ingestion

{
  "url": "https://internal-origin.example.com/feed",
  "resolve": {
    "internal-origin.example.com": "192.0.2.42"
  },
  "redirect": "manual"
}

Impersonation Presets & Fingerprints

@trishchuk/fetch MCP supports a wide range of browser profiles:

Note: the underlying client, @trishchuk/fetch, ships as native prebuilt binaries per-platform. If you're on an unsupported platform, run the server on a supported host or via a container.

Further Reading