plasmate-a11y

March 26, 2026 · View on GitHub

Accessibility auditing powered by Plasmate's Semantic Object Model.

The insight: SOM is what screen readers always wanted — structured content with clear hierarchy and meaningful labels. By analyzing the SOM, we can instantly identify what a screen reader would "see" vs what's actually on the page.

Install

pip install plasmate plasmate-a11y

Quick Start

from plasmate_a11y import audit_url

result = audit_url("https://example.com")
print(f"Score: {result['summary']['score']}/100")
print(f"Errors: {result['summary']['errors']}")
print(f"Warnings: {result['summary']['warnings']}")

for issue in result['issues']:
    print(f"  [{issue['severity']}] {issue['message']}")

CLI

python -m plasmate_a11y https://example.com

Exit code is 1 if any errors are found — useful in CI pipelines.

What It Checks

Rule	Severity	Description
`no-title`	Error	Page missing title element
`no-lang`	Error	Page missing language attribute
`heading-skip`	Warning	Heading levels skip (e.g. h2 → h4)
`no-h1`	Error	Page doesn't start with h1
`img-no-alt`	Error	Images without alt text
`empty-link-text`	Error	Links with no text
`vague-link-text`	Warning	Links with "click here" etc.
`input-no-label`	Error	Form inputs without labels
`no-nav-landmark`	Warning	Missing navigation region
`no-main-landmark`	Warning	Missing main content region

Programmatic Use

from plasmate_a11y import audit_url, audit_urls, format_report, format_markdown

# Single URL
result = audit_url("https://example.com")
print(format_report(result))          # terminal output with color
print(format_markdown(result))        # markdown table

# Multiple URLs
results = audit_urls(["https://a.com", "https://b.com"])
for r in results:
    print(format_report(r))

Why Plasmate for Accessibility?

Traditional accessibility auditors use Chrome + axe-core, which requires rendering the full page. Plasmate skips rendering and goes straight to semantic analysis:

10× faster — no rendering overhead
Structured by default — SOM already represents what matters for a11y
Token-efficient — results are compact and machine-readable

Development

git clone https://github.com/plasmate-labs/plasmate-a11y.git
cd plasmate-a11y
pip install -e ".[dev]"
pytest

License

Apache 2.0 — see LICENSE.

Part of the Plasmate Ecosystem


Engine	plasmate - The browser engine for agents
MCP	plasmate-mcp - Claude Code, Cursor, Windsurf
Extension	plasmate-extension - Chrome cookie export
SDKs	Python / Node.js / Go / Rust
Frameworks	LangChain / CrewAI / AutoGen / Smolagents
Tools	Scrapy / Audit / A11y / GitHub Action
Resources	Awesome Plasmate / Notebooks / Benchmarks
Docs	docs.plasmate.app
W3C	Web Content Browser for AI Agents