plasmate-audit

March 26, 2026 · View on GitHub

Site auditing powered by Plasmate - 10x faster than Chrome-based tools.

Install

pip install plasmate plasmate-audit

Quick Start

# Audit a site (crawls up to 50 pages)
plasmate-audit https://example.com

# More pages
plasmate-audit https://example.com --max-pages 200

# JSON output
plasmate-audit https://example.com --json

Python API

from plasmate_audit import audit_site

result = audit_site("https://example.com", max_pages=50)
print(f"Score: {result['score']}/100")
print(f"Pages: {result['pages_crawled']}")
print(f"Issues: {result['errors']} errors, {result['warnings']} warnings")

What It Checks

CheckSeverityDescription
missing-titleErrorPage has no title
title-too-longWarningTitle >60 characters
title-too-shortWarningTitle <10 characters
no-h1ErrorNo H1 heading
multiple-h1WarningMore than one H1
no-headingsWarningNo headings at all
empty-linksWarningLinks with no text
thin-contentWarningPages with <100 chars of content
images-no-altErrorImages without alt text

Why Plasmate?

Screaming Frog and Lighthouse use Chrome to render every page. Plasmate skips rendering:

ToolSpeed (100 pages)Memory
Screaming Frog~5 minutes2GB+
Lighthouse~15 minutes500MB+
plasmate-audit~30 seconds50MB

How It Works

  1. Crawl — Start from a URL, discover internal links via Plasmate's SOM
  2. Analyze — Run SEO/content checks against each page's structured object model
  3. Report — Get a score, issue list, and per-page breakdown

No headless browser. No JavaScript execution. Just fast, structured analysis.

License

Apache 2.0


Part of the Plasmate Ecosystem

Engineplasmate - The browser engine for agents
MCPplasmate-mcp - Claude Code, Cursor, Windsurf
Extensionplasmate-extension - Chrome cookie export
SDKsPython / Node.js / Go / Rust
FrameworksLangChain / CrewAI / AutoGen / Smolagents
ToolsScrapy / Audit / A11y / GitHub Action
ResourcesAwesome Plasmate / Notebooks / Benchmarks
Docsdocs.plasmate.app
W3CWeb Content Browser for AI Agents