Skip to content

CLI Reference

Commands

agent-a11y audit

Run an accessibility audit.

agent-a11y audit <url> [flags]

Arguments:

  • <url> - URL to audit (required)

Flags:

Flag Short Description Default
--level WCAG level (A, AA, AAA) AA
--version WCAG version (2.0, 2.1, 2.2) 2.2
--format -f Output format json
--output -o Output file path stdout
--human Output human-readable format false
--design-system Path to design system spec
--source-map Path to source map directory
--crawl Enable site crawling false
--depth Crawl depth 2
--max-pages Maximum pages 50
--delay Request delay 500ms
--journey Journey file path
--specialized Run specialized tests false
--llm-provider LLM provider
--llm-model LLM model
--headless Headless browser true
--timeout Browser timeout 30s

Examples:

# Basic audit (agent-optimized JSON output)
agent-a11y audit https://example.com

# Human-readable output
agent-a11y audit https://example.com --human

# With design system token suggestions
agent-a11y audit https://example.com --design-system ./design-system/

# With source mapping (maps findings to source files)
agent-a11y audit https://localhost:3000 --source-map ./dist/

# WCAG 2.1 Level AAA
agent-a11y audit https://example.com --level AAA --version 2.1

# Site crawl with HTML report
agent-a11y audit https://example.com --crawl -o report.html -f html

# With LLM evaluation
agent-a11y audit https://example.com --llm-provider anthropic --llm-model claude-sonnet-4-20250514

agent-a11y serve

Start the HTTP API server.

agent-a11y serve [flags]

Flags:

Flag Description Default
--port Server port 8080
--host Server host localhost

agent-a11y mcp serve

Start the MCP server for AI assistant integration.

agent-a11y mcp serve

Communicates via stdio for MCP protocol.

agent-a11y compare

Compare accessibility between two URLs (VPAT comparison).

agent-a11y compare <before-url> <after-url> [flags]

Flags:

Flag Description Default
--name Comparison name URL-based
--from-files Compare from existing JSON files false
--format Output format (json, html, markdown, vpat) markdown
--output Output file path stdout

Shows:

  • Issues fixed
  • Issues introduced
  • Issues unchanged

agent-a11y validate

Validate accessibility fixes against a baseline. Enables the autonomous fix loop.

agent-a11y validate <url> [flags]

Flags:

Flag Short Description Default
--baseline -b Baseline audit file to compare against
--expect-fixed Rule IDs expected to be fixed (comma-separated)
--fail-on-regression Exit with code 1 if regressions detected false
--save-baseline Save this audit as baseline (no comparison) false
--include-results Include full before/after results in delta false
--output -o Output file path stdout

Examples:

# Create baseline from first audit
agent-a11y validate https://example.com --save-baseline -o baseline.json

# Validate after fixes
agent-a11y validate https://example.com --baseline baseline.json

# Check specific rules were fixed
agent-a11y validate https://example.com --baseline baseline.json \
  --expect-fixed color-contrast,image-alt

# CI mode: fail on any regressions
agent-a11y validate https://example.com --baseline baseline.json \
  --fail-on-regression

Output:

The validate command outputs a ValidationDelta JSON with:

  • fixed: Issues that were resolved
  • remaining: Issues still present
  • regressions: New issues introduced
  • status: FIXED, IMPROVED, NO_CHANGE, REGRESSED, or MIXED
  • improvement: Percentage change

agent-a11y config

Manage project-specific fix configuration.

agent-a11y config <subcommand>

Subcommands:

config show

Show resolved configuration for the current project.

agent-a11y config show

Displays:

  • Detected language and framework
  • Component library (MUI, Chakra, shadcn, etc.)
  • Matched project configuration
  • Config file location

config init

Create a default configuration file.

agent-a11y config init

Creates ~/.plexusone/a11y/fixes.yaml with example configuration.

agent-a11y demo

Run demo audits on sample sites.

# List available demos
agent-a11y demo list

# Run a specific demo
agent-a11y demo run w3c-bad

# Generate a demo site
agent-a11y demo generate my-demo

agent-a11y version

Show version information.

agent-a11y version

Global Flags

Flag Description
--config Config file path
--verbose Verbose output
--help Show help

Exit Codes

Code Description
0 Success (conformant)
1 Non-conformant (issues found)
2 Error (audit failed)