webclaw/README.md

The fastest web scraper for AI agents.
_{67% fewer tokens. Sub-millisecond extraction. Zero browser overhead.}

---

_{Claude Code's built-in web_fetch → 403 Forbidden. noxa → clean markdown.}

---

Your AI agent calls fetch() and gets a 403. Or 142KB of raw HTML that burns through your token budget. noxa fixes both.

It extracts clean, structured content from any URL using Chrome-level TLS fingerprinting — no headless browser, no Selenium, no Puppeteer. Output is optimized for LLMs: 67% fewer tokens than raw HTML, with metadata, links, and images preserved.

                     Raw HTML                          noxa
┌──────────────────────────────────┐    ┌──────────────────────────────────┐
│ <div class="ad-wrapper">         │    │ # Breaking: AI Breakthrough      │
│ <nav class="global-nav">         │    │                                  │
│ <script>window.__NEXT_DATA__     │    │ Researchers achieved 94%         │
│ ={...8KB of JSON...}</script>    │    │ accuracy on cross-domain         │
│ <div class="social-share">       │    │ reasoning benchmarks.            │
│ <button>Tweet</button>           │    │                                  │
│ <footer class="site-footer">     │    │ ## Key Findings                  │
│ <!-- 142,847 characters -->      │    │ - 3x faster inference            │
│                                  │    │ - Open-source weights            │
│         4,820 tokens             │    │         1,590 tokens             │
└──────────────────────────────────┘    └──────────────────────────────────┘

---

Get Started (30 seconds)

For AI agents (Claude, Cursor, Windsurf, VS Code)

npx create-noxa

Auto-detects your AI tools, downloads the MCP server, and configures everything. One command.

Homebrew (macOS/Linux)

brew tap jmagar/noxa
brew install noxa

Prebuilt binaries

Download from GitHub Releases for macOS (arm64, x86_64) and Linux (x86_64, aarch64).

Cargo (from source)

cargo install --git https://github.com/jmagar/noxa.git noxa-cli --bin noxa
cargo install --git https://github.com/jmagar/noxa.git noxa-mcp

Docker

docker run --rm ghcr.io/0xmassi/noxa https://example.com

Docker Compose (with Ollama for LLM features)

cp env.example .env
docker compose up -d

---

Why noxa?

	noxa	Firecrawl	Trafilatura	Readability
Extraction accuracy	95.1%	—	80.6%	83.5%
Token efficiency	-67%	—	-55%	-51%
Speed (100KB page)	3.2ms	~500ms	18.4ms	8.7ms
TLS fingerprinting	Yes	No	No	No
Self-hosted	Yes	No	Yes	Yes
MCP (Claude/Cursor)	Yes	No	No	No
No browser required	Yes	No	Yes	Yes
Cost	Free		Free	Free

Choose noxa if you want fast local extraction, LLM-optimized output, and native AI agent integration.

---

What it looks like

$ noxa https://stripe.com -f llm

> URL: https://stripe.com
> Title: Stripe | Financial Infrastructure for the Internet
> Language: en
> Word count: 847

# Stripe | Financial Infrastructure for the Internet

Stripe is a suite of APIs powering online payment processing
and commerce solutions for internet businesses of all sizes.

## Products
- Payments — Accept payments online and in person
- Billing — Manage subscriptions and invoicing
- Connect — Build a marketplace or platform
...

$ noxa https://github.com --brand

{
  "name": "GitHub",
  "colors": [{"hex": "#59636E", "usage": "Primary"}, ...],
  "fonts": ["Mona Sans", "ui-monospace"],
  "logos": [{"url": "https://github.githubassets.com/...", "kind": "svg"}]
}

$ noxa https://docs.rust-lang.org --crawl --depth 2 --max-pages 50

Crawling... 50/50 pages extracted
---
# Page 1: https://docs.rust-lang.org/
...
# Page 2: https://docs.rust-lang.org/book/
...

---

Examples

Basic Extraction

# Extract as markdown (default)
noxa https://example.com

# Multiple output formats
noxa https://example.com -f markdown    # Clean markdown
noxa https://example.com -f json        # Full structured JSON
noxa https://example.com -f text        # Plain text (no formatting)
noxa https://example.com -f llm         # Token-optimized for LLMs (67% fewer tokens)

# Bare domains work (auto-prepends https://)
noxa example.com

Content Filtering

# Only extract main content (skip nav, sidebar, footer)
noxa https://docs.rs/tokio --only-main-content

# Include specific CSS selectors
noxa https://news.ycombinator.com --include ".titleline,.score"

# Exclude specific elements
noxa https://example.com --exclude "nav,footer,.ads,.sidebar"

# Combine both
noxa https://docs.rs/reqwest --only-main-content --exclude ".sidebar"

Brand Identity Extraction

# Extract colors, fonts, logos from any website
noxa --brand https://stripe.com
# Output: { "name": "Stripe", "colors": [...], "fonts": ["Sohne"], "logos": [...] }

noxa --brand https://github.com
# Output: { "name": "GitHub", "colors": [{"hex": "#1F2328", ...}], "fonts": ["Mona Sans"], ... }

noxa --brand wikipedia.org
# Output: 10 colors, 5 fonts, favicon, logo URL

Sitemap Discovery

# Discover all URLs from a site's sitemaps
noxa --map https://sitemaps.org
# Output: one URL per line (84 URLs found)

# JSON output with metadata
noxa --map https://sitemaps.org -f json
# Output: [{ "url": "...", "last_modified": "...", "priority": 0.8 }]

Recursive Crawling

# Crawl a site (default: depth 1, max 20 pages)
noxa --crawl https://example.com

# Control depth and page limit
noxa --crawl --depth 2 --max-pages 50 https://docs.rs/tokio

# Crawl with sitemap seeding (finds more pages)
noxa --crawl --sitemap --depth 2 https://docs.rs/tokio

# Filter crawl paths
noxa --crawl --include-paths "/api/*,/guide/*" https://docs.example.com
noxa --crawl --exclude-paths "/changelog/*,/blog/*" https://docs.example.com

# Control concurrency and delay
noxa --crawl --concurrency 10 --delay 200 https://example.com

Change Detection (Diff)

# Step 1: Save a snapshot
noxa https://example.com -f json > snapshot.json

# Step 2: Later, compare against the snapshot
noxa --diff-with snapshot.json https://example.com
# Output:
#   Status: Same
#   Word count delta: +0

# If the page changed:
#   Status: Changed
#   Word count delta: +42
#   --- old
#   +++ new
#   @@ -1,3 +1,3 @@
#   -Old content here
#   +New content here

PDF Extraction

# PDF URLs are auto-detected via Content-Type
noxa https://example.com/report.pdf

# Control PDF mode
noxa --pdf-mode auto https://example.com/report.pdf  # Error on empty (catches scanned PDFs)
noxa --pdf-mode fast https://example.com/report.pdf  # Return whatever text is found

Batch Processing

# Multiple URLs in one command
noxa https://example.com https://httpbin.org/html https://rust-lang.org

# URLs from a file (one per line, # comments supported)
noxa --urls-file urls.txt

# Batch with JSON output
noxa --urls-file urls.txt -f json

# Proxy rotation for large batches
noxa --urls-file urls.txt --proxy-file proxies.txt --concurrency 10

Local Files & Stdin

# Extract from a local HTML file
noxa --file page.html

# Pipe HTML from another command
curl -s https://example.com | noxa --stdin

# Chain with other tools
noxa https://example.com -f text | wc -w    # Word count
noxa https://example.com -f json | jq '.metadata.title'  # Extract title with jq

Browser Impersonation

# Chrome (default) — latest Chrome TLS fingerprint
noxa https://example.com

# Firefox fingerprint
noxa --browser firefox https://example.com

# Random browser per request (good for batch)
noxa --browser random --urls-file urls.txt

Custom Headers & Cookies

# Custom headers
noxa -H "Authorization: Bearer token123" https://api.example.com
noxa -H "Accept-Language: de-DE" https://example.com

# Cookies
noxa --cookie "session=abc123; theme=dark" https://example.com

# Multiple headers
noxa -H "X-Custom: value" -H "Authorization: Bearer token" https://example.com

LLM-Powered Features

These require an LLM provider (Ollama local, or OpenAI/Anthropic API key).

# Summarize a page (default: 3 sentences)
noxa --summarize https://example.com

# Control summary length
noxa --summarize 5 https://example.com

# Extract structured JSON with a schema
noxa --extract-json '{"type":"object","properties":{"title":{"type":"string"},"price":{"type":"number"}}}' https://example.com/product

# Extract with a schema from file
noxa --extract-json @schema.json https://example.com/product

# Extract with natural language prompt
noxa --extract-prompt "Get all pricing tiers with name, price, and features" https://stripe.com/pricing

# Use a specific LLM provider
noxa --llm-provider ollama --summarize https://example.com
noxa --llm-provider openai --llm-model gpt-4o --extract-prompt "..." https://example.com
noxa --llm-provider anthropic --summarize https://example.com

Raw HTML Output

# Get the raw fetched HTML (no extraction)
noxa --raw-html https://example.com

# Useful for debugging extraction issues
noxa --raw-html https://example.com > raw.html
noxa --file raw.html  # Then extract locally

Metadata & Verbose Mode

# Include YAML frontmatter with metadata
noxa --metadata https://example.com
# Output:
#   ---
#   title: "Example Domain"
#   source: "https://example.com"
#   word_count: 20
#   ---
#   # Example Domain
#   ...

# Verbose logging (debug extraction pipeline)
noxa -v https://example.com

Proxy Usage

# Single proxy
noxa --proxy http://user:pass@proxy.example.com:8080 https://example.com

# SOCKS5 proxy
noxa --proxy socks5://proxy.example.com:1080 https://example.com

# Proxy rotation from file (one per line: host:port:user:pass)
noxa --proxy-file proxies.txt https://example.com

# Auto-load proxies.txt from current directory
echo "proxy1.com:8080:user:pass" > proxies.txt
noxa https://example.com  # Automatically detects and uses proxies.txt

Real-World Recipes

# Monitor competitor pricing — save today's pricing
noxa --extract-json '{"type":"array","items":{"type":"object","properties":{"plan":{"type":"string"},"price":{"type":"string"}}}}' \
  https://competitor.com/pricing -f json > pricing-$(date +%Y%m%d).json

# Build a documentation search index
noxa --crawl --sitemap --depth 3 --max-pages 500 -f llm https://docs.example.com > docs.txt

# Extract all images from a page
noxa https://example.com -f json | jq -r '.content.images[].src'

# Get all external links
noxa https://example.com -f json | jq -r '.content.links[] | select(.href | startswith("http")) | .href'

# Compare two pages
noxa https://site-a.com -f json > a.json
noxa https://site-b.com --diff-with a.json

---

MCP Server — 10 tools for AI agents

noxa ships as an MCP server that plugs into Claude Desktop, Claude Code, Cursor, Windsurf, OpenCode, Antigravity, Codex CLI, and any MCP-compatible client.

npx create-noxa    # auto-detects and configures everything

Or manual setup — add to your Claude Desktop config:

{
  "mcpServers": {
    "noxa": {
      "command": "~/.noxa/noxa-mcp"
    }
  }
}

Then in Claude: "Scrape the top 5 results for 'web scraping tools' and compare their pricing" — it just works.

Available tools

Tool	Description	Requires API key?
scrape	Extract content from any URL	No
crawl	Recursive site crawl	No
map	Discover URLs from sitemaps	No
batch	Parallel multi-URL extraction	No
extract	LLM-powered structured extraction	No (needs Ollama)
summarize	Page summarization	No (needs Ollama)
diff	Content change detection	No
brand	Brand identity extraction	No
search	Web search + scrape results	Yes
research	Deep multi-source research	Yes

8 of 10 tools work locally — no account, no API key, fully private.

---

Features

Extraction

• Readability scoring — multi-signal content detection (text density, semantic tags, link ratio)
• Noise filtering — strips nav, footer, ads, modals, cookie banners (Tailwind-safe)
• Data island extraction — catches React/Next.js JSON payloads, JSON-LD, hydration data
• YouTube metadata — structured data from any YouTube video
• PDF extraction — auto-detected via Content-Type
• 5 output formats — markdown, text, JSON, LLM-optimized, HTML

Content control

noxa URL --include "article, .content"       # CSS selector include
noxa URL --exclude "nav, footer, .sidebar"    # CSS selector exclude
noxa URL --only-main-content                  # Auto-detect main content

Crawling

noxa URL --crawl --depth 3 --max-pages 100   # BFS same-origin crawl
noxa URL --crawl --sitemap                    # Seed from sitemap
noxa URL --map                                # Discover URLs only

LLM features (Ollama / OpenAI / Anthropic)

noxa URL --summarize                          # Page summary
noxa URL --extract-prompt "Get all prices"    # Natural language extraction
noxa URL --extract-json '{"type":"object"}'   # Schema-enforced extraction

Change tracking

noxa URL -f json > snap.json                  # Take snapshot
noxa URL --diff-with snap.json                # Compare later

Brand extraction

noxa URL --brand                              # Colors, fonts, logos, OG image

Proxy rotation

noxa URL --proxy http://user:pass@host:port   # Single proxy
noxa URLs --proxy-file proxies.txt            # Pool rotation

---

Benchmarks

All numbers from real tests on 50 diverse pages. See benchmarks/ for methodology and reproduction instructions.

Extraction quality

Accuracy      noxa     ███████████████████ 95.1%
              readability ████████████████▋   83.5%
              trafilatura ████████████████    80.6%
              newspaper3k █████████████▎      66.4%

Noise removal noxa     ███████████████████ 96.1%
              readability █████████████████▊  89.4%
              trafilatura ██████████████████▏ 91.2%
              newspaper3k ███████████████▎    76.8%

Speed (pure extraction, no network)

10KB page     noxa     ██                   0.8ms
              readability █████                2.1ms
              trafilatura ██████████           4.3ms

100KB page    noxa     ██                   3.2ms
              readability █████                8.7ms
              trafilatura ██████████           18.4ms

Token efficiency (feeding to Claude/GPT)

Format	Tokens	vs Raw HTML
Raw HTML	4,820	baseline
readability	2,340	-51%
trafilatura	2,180	-55%
noxa llm	1,590	-67%

Crawl speed

Concurrency	noxa	Crawl4AI	Scrapy
5	9.8 pg/s	5.2 pg/s	7.1 pg/s
10	18.4 pg/s	8.7 pg/s	12.3 pg/s
20	32.1 pg/s	14.2 pg/s	21.8 pg/s

---

Architecture

noxa/
  crates/
    noxa-core     Pure extraction engine. Zero network deps. WASM-safe.
    noxa-fetch    HTTP client + TLS fingerprinting (wreq/BoringSSL). Crawler. Batch ops.
    noxa-llm      LLM provider chain (Ollama -> OpenAI -> Anthropic)
    noxa-pdf      PDF text extraction
    noxa-mcp      MCP server (10 tools for AI agents)
    noxa      CLI binary

noxa-core takes raw HTML as a &str and returns structured output. No I/O, no network, no allocator tricks. Can compile to WASM.

---

Configuration

Non-secret defaults live in config.json in your working directory. Copy the example:

cp config.example.json config.json

Precedence: CLI flags > config.json > built-in defaults

Secrets and URLs (API keys, proxy, webhook, LLM base URL) always go in .env, not config.json:

cp env.example .env

Override config path for a single run:

NOXA_CONFIG=/path/to/other-config.json noxa https://example.com
NOXA_CONFIG=/dev/null noxa https://example.com  # bypass config entirely

Bool flag limitation: flags like --metadata, --only-main-content, --verbose set to true in config.json cannot be overridden to false from the CLI for a single run (clap has no --no-flag variant). Use NOXA_CONFIG=/dev/null to bypass.

Environment variables

Variable	Description
NOXA_API_KEY	Cloud API key (enables bot bypass, JS rendering, search, research)
OLLAMA_HOST	Ollama URL for local LLM features (default: http://localhost:11434)
OPENAI_API_KEY	OpenAI API key for LLM features
ANTHROPIC_API_KEY	Anthropic API key for LLM features
NOXA_PROXY	Single proxy URL
NOXA_PROXY_FILE	Path to proxy pool file

---

Cloud API (optional)

For bot-protected sites, JS rendering, and advanced features, noxa offers a hosted API at noxa.io.

The CLI and MCP server work locally first. Cloud is used as a fallback when:

• A site has bot protection (Cloudflare, DataDome, WAF)
• A page requires JavaScript rendering
• You use search or research tools

export NOXA_API_KEY=wc_your_key

# Automatic: tries local first, cloud on bot detection
noxa https://protected-site.com

# Force cloud
noxa --cloud https://spa-site.com

SDKs

npm install @noxa/sdk                  # TypeScript/JavaScript
pip install noxa                        # Python
go get github.com/jmagar/noxa-go      # Go

---

Use cases

• AI agents — Give Claude/Cursor/GPT real-time web access via MCP
• Research — Crawl documentation, competitor sites, news archives
• Price monitoring — Track changes with --diff-with snapshots
• Training data — Prepare web content for fine-tuning with token-optimized output
• Content pipelines — Batch extract + summarize in CI/CD
• Brand intelligence — Extract visual identity from any website

---

Community

• GitHub Issues — bug reports and feature requests

Contributing

We welcome contributions! See CONTRIBUTING.md for guidelines.

• Good first issues
• Architecture docs

Acknowledgments

TLS and HTTP/2 browser fingerprinting is powered by wreq and http2 by @0x676e67, who pioneered browser-grade HTTP/2 fingerprinting in Rust.

License

AGPL-3.0