servo-fetch 0.1.0

servo-fetch is a single-binary CLI and MCP server that renders web pages using the Servo browser engine. It executes JavaScript, computes CSS layout, captures screenshots, and extracts clean content — all without downloading a browser.

servo-fetch "https://example.com"                        # Clean Markdown
servo-fetch "https://example.com" --screenshot page.png  # PNG screenshot, no GPU needed
servo-fetch "https://example.com" --js "document.title"  # Run JS in the page

Why servo-fetch?

• Screenshots without a browser runtime. Servo renders pages to PNG with a software renderer. No GPU, no Xvfb, no Chromium download. Drop the binary into Docker or CI and it just works.

• Reads JavaScript-heavy pages. SPAs, React, Vue — servo-fetch executes JS via SpiderMonkey and extracts the rendered content. Plain HTTP fetchers return empty HTML; servo-fetch returns the real page.

• Strips navigation noise using CSS layout. Most tools guess page structure from HTML tags. servo-fetch uses getComputedStyle() and getBoundingClientRect() to detect fixed navbars, sidebars, and footers — then removes them before extraction.

• Single binary, zero runtime dependencies. cargo install or download a prebuilt binary. No Node.js, no npx playwright install, no apt-get install chromium.

• Built-in MCP server for AI agents. Three tools (fetch, screenshot, execute_js) over stdio or Streamable HTTP. AI agents can read SPAs and take screenshots without any browser setup.

Install

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

Or download from GitHub Releases, or build from source (requires Rust 1.86.0+):

cargo binstall servo-fetch   # prebuilt binary
cargo install servo-fetch    # build from source

Usage

# Readable Markdown (default)
servo-fetch "https://example.com"

# Structured JSON
servo-fetch "https://example.com" --json

# Screenshot — rendered to PNG without GPU
servo-fetch "https://example.com" --screenshot page.png

# Execute JavaScript in the page context
servo-fetch "https://example.com" --js "document.title"

# Extract a specific section by CSS selector
servo-fetch "https://example.com" --selector "article"

# PDF text extraction (auto-detected)
servo-fetch "https://example.com/report.pdf"

Options

JSON output

--json returns an object with these fields:

Fields marked ? are omitted when not detected.

How it works

1. Servo loads the page and executes JavaScript via SpiderMonkey
2. CSS is computed with Servo's parallel layout engine — getComputedStyle() and getBoundingClientRect() identify page structure
3. Navbars, sidebars, and footers are stripped using CSS layout data
4. Mozilla's Readability algorithm extracts the main content
5. Content is output as Markdown, JSON, or PNG

PDF URLs are auto-detected via Content-Type and extracted directly without Servo.

MCP server

servo-fetch includes a built-in MCP server for AI agents with three tools: fetch, screenshot, and execute_js.

# stdio transport (default)
servo-fetch mcp

# Streamable HTTP transport
servo-fetch mcp --port 8080

Add to your MCP client config (Claude Code, Codex, Cursor, etc.):

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

Security

servo-fetch blocks all private and reserved IP ranges (RFC 6890), strips credentials from URLs, validates redirect targets, and sanitizes all output against terminal escape injection (CVE-2021-42574). See SECURITY.md for details.

Limitations

• Best suited for documentation, blogs, and SSR sites
• Some SPAs with complex client-side rendering may not fully render
• Servo's web compatibility is improving monthly

License

MIT