v0.1.0¶
Release Date: 2026-02-28
Highlights¶
- Go Accessibility Toolkit - Comprehensive WCAG compliance testing for Go applications
- axe-core + LLM - Industry-standard rules enhanced with AI evaluation to reduce false positives
- Multiple Audit Modes - Single page, site crawling, and user journey testing
- Rich Report Formats - JSON, HTML, Markdown, VPAT 2.4, WCAG-EM, CSV
Overview¶
agent-a11y is a Go accessibility auditing toolkit for WCAG 2.0, 2.1, and 2.2 compliance testing. It combines axe-core rules with LLM-as-a-Judge evaluation to provide accurate accessibility assessments with reduced false positives.
Features¶
WCAG Compliance Testing¶
- Support for WCAG 2.0, 2.1, and 2.2
- Levels A, AA, and AAA conformance testing
- Comprehensive success criteria mapping
Audit Modes¶
| Mode | Description | Use Case |
|---|---|---|
| Single Page | Audit one URL | Quick checks, CI pipelines |
| Site Crawl | Crawl and audit multiple pages | Full site audits |
| Journey | Follow user journeys via YAML | Authenticated flows, SPAs |
axe-core Integration¶
- Industry-standard accessibility rules
- Automatic rule updates
- Detailed issue descriptions and help links
LLM-as-a-Judge¶
Reduce false positives with AI evaluation:
- Providers: Anthropic, OpenAI, Google, xAI, Ollama
- Evaluation: Confirms findings and provides reasoning
- Confidence: Scores for prioritization
Specialized Tests¶
Browser-based tests beyond axe-core:
- Keyboard navigation and trap detection
- Focus visibility and order
- Reflow at 400% zoom
- Touch target sizing
- Text spacing override support
- Hover content accessibility
- Flash/flicker detection
Report Formats¶
| Format | Description |
|---|---|
| JSON | Machine-readable for CI/CD |
| HTML | Interactive reports for stakeholders |
| Markdown | Documentation and version control |
| VPAT 2.4 | Procurement compliance documentation |
| WCAG-EM | W3C evaluation methodology format |
| CSV | Spreadsheet export |
Integration Options¶
- CLI - Command-line tool for scripts and CI/CD
- HTTP API - REST API at
/auditand/health - MCP Server - Integration with AI assistants (Claude)
- Go Library - Embed in Go applications
- Multi-Agent Spec - Use in multi-agent workflows
Installation¶
Quick Start¶
# Basic audit
agent-a11y audit https://example.com
# WCAG 2.2 Level AA with HTML report
agent-a11y audit https://example.com --level AA --version 2.2 -o report.html -f html
# With LLM evaluation
agent-a11y audit https://example.com --llm-provider anthropic --llm-model claude-sonnet-4-20250514
Go Library¶
auditor, _ := a11y.New(
a11y.WithLevel(a11y.LevelAA),
a11y.WithVersion(a11y.Version22),
)
defer auditor.Close()
result, _ := auditor.AuditPage(ctx, "https://example.com")
fmt.Println(result.Summary())
Dependencies¶
- vibium-go v0.2.1 - Browser automation
- omnillm v0.13.0 - Multi-provider LLM support
- multi-agent-spec v0.8.0 - Multi-agent workflows
- MCP SDK v1.4.0 - Model Context Protocol