servo-fetch-cli

A browser engine in a binary — fetch, render, and extract web content powered by Servo.

For programmatic use in Rust, see the servo-fetch library crate.

Install

Pre-built binaries (recommended)

curl -fsSL https://raw.githubusercontent.com/konippi/servo-fetch/main/install.sh | sh

Cargo

cargo binstall servo-fetch-cli   # prebuilt binary via cargo-binstall
cargo install servo-fetch-cli    # build from source (requires Rust 1.86.0+)

Usage

Extract content

servo-fetch "https://example.com"              # Readable Markdown (default)
servo-fetch "https://example.com" --json       # Structured JSON
servo-fetch "https://example.com" --raw html   # Raw rendered HTML
servo-fetch "https://example.com" --raw text   # Plain text (innerText)

Batch fetch

servo-fetch URL1 URL2 URL3                     # Parallel fetch, Markdown output
servo-fetch URL1 URL2 --json                   # Parallel fetch, NDJSON output

Screenshots

servo-fetch "https://example.com" --screenshot page.png
servo-fetch "https://example.com" --screenshot full.png --full-page

JavaScript execution

servo-fetch "https://example.com" --js "document.title"
servo-fetch "https://example.com" --js "document.querySelectorAll('h2').length"

CSS selector extraction

servo-fetch "https://example.com" --selector "article"
servo-fetch "https://example.com" --selector ".main-content" --json

Crawl a site

servo-fetch crawl "https://docs.example.com" --limit 20
servo-fetch crawl "https://docs.example.com" --include "/docs/**" --exclude "/docs/archive/**"
servo-fetch crawl "https://docs.example.com" --json --max-depth 5

SPA / dynamic content

servo-fetch "https://spa.example.com" --settle 3000       # Wait 3s after load for hydration
servo-fetch "https://spa.example.com" -t 60 --settle 5000 # 60s timeout + 5s settle

MCP server

servo-fetch mcp                # stdio transport (for AI agents)
servo-fetch mcp --port 8080    # Streamable HTTP transport

Options

JSON output

--json returns an object with these fields:

Crawl subcommand

servo-fetch crawl <URL> follows same-site links using BFS. Respects robots.txt (RFC 9309) with a minimum 500ms interval.

Logging

Diagnostic messages go to stderr; stdout is reserved for data output so pipes stay clean.

servo-fetch -v "https://example.com"                       # info and above
servo-fetch -vv "https://example.com"                      # debug
servo-fetch -vvv "https://example.com"                     # trace
servo-fetch -q "https://example.com"                       # errors only
RUST_LOG="servo_fetch=debug" servo-fetch "https://..."     # fine-grained override
RUST_LOG="servo_fetch=trace,servo=debug" servo-fetch "..." # include Servo internals

RUST_LOG uses tracing-subscriber's directive syntax and always wins over CLI flags.

Environment Variables

MCP Server

Built-in Model Context Protocol server over stdio or Streamable HTTP.

{
  "mcpServers": {
    "servo-fetch": {
      "command": "servo-fetch",
      "args": ["mcp"]
    }
  }
}

Streamable HTTP: servo-fetch mcp --port 8080