Document Accessibility Parameter Reference

This document provides a comprehensive reference for all parameters used in the Document Accessibility tool, ensuring consistent usage between the command-line interface (CLI) and the API.

Standardized Parameter Names

The following table shows the standardized parameter names used across the tool:

Parameter	CLI Option	API Parameter	Required	Default	Description
Common Parameters
Input Path	--input, -i	input_path, pdf_path, html_path	Yes	N/A	Path to input file or directory
Output Path	--output, -o	output_dir, output_path	No	Auto-generated based on input	Path for output files or directory
Debug Mode	--debug	N/A	No	Disabled	Enable debug logging
Quiet Mode	--quiet, -q	quiet	No	Disabled	Suppress non-essential output
Config File	--config, -c	config_path	No	None	Path to configuration file
Save Config	--save-config	N/A	No	None	Save current configuration to file
AWS Profile	--profile	profile	No	Default AWS profile	AWS profile name to use
PDF Conversion
S3 Bucket	--s3-bucket	s3_bucket	Yes (for PDF processing)	None	Name of S3 bucket for storing input/output files
BDA Project ARN	--bda-project-arn	bda_project_arn	No	None	ARN of an existing BDA project
Create BDA Project	--create-bda-project	create_bda_project	No	Disabled	Create a new BDA project if needed
Format	--format, -f	format	No	html	Output format (html, md)
Single File	--single-file	single_file	No	Disabled	Generate a single HTML file
Single Page	--single-page	single_page	No	Disabled	Combine pages into one document
Multi Page	--multi-page	multi_page	No	Disabled	Keep pages as separate files
Continuous	--continuous	continuous	No	Disabled	Use continuous scrolling
Extract Images	--extract-images	extract_images	No	Enabled	Extract images from PDF
Image Format	--image-format	image_format	No	png	Format for extracted images (png, jpg, webp)
Embed Images	--embed-images	embed_images	No	Disabled	Embed images as data URIs
Exclude Images	--exclude-images	exclude_images	No	Disabled	Don’t include images
Accessibility Audit
Format	--format, -f	report_format	No	json	Output format for report (json, html, text)
Checks	--checks	issue_types	No	All checks	Comma-separated list of checks to run
Severity	--severity	severity_threshold	No	minor	Minimum severity level (minor, major, critical)
Detailed	--detailed	detailed, include_context	No	Enabled	Include detailed context information
Summary Only	--summary-only	summary_only	No	Disabled	Only include summary information
Accessibility Remediation
Auto Fix	--auto-fix	auto_fix	No	Disabled	Automatically fix issues where possible
Max Issues	--max-issues	max_issues	No	All issues	Maximum number of issues to remediate
Model ID	--model-id	model_id	No	Default model	Bedrock model ID to use for remediation
Severity Threshold	--severity-threshold	severity_threshold	No	minor	Minimum severity level to remediate
Audit Report	--audit-report	audit_report	No	None	Path to audit report JSON file
Generate Report	--generate-report	generate_report	No	Enabled	Generate a remediation report
Report Format	--report-format	report_format	No	html	Format for the remediation report
Process Command
Skip Audit	--skip-audit	perform_audit=False	No	Disabled	Skip the audit step
Skip Remediation	--skip-remediation	perform_remediation=False	No	Disabled	Skip the remediation step
Audit Format	--audit-format	audit_format	No	json	Format for the audit report

CLI Usage Examples

Convert a PDF to HTML

content-accessibility-utility convert --input document.pdf --output output_dir/ --single-file --extract-images

Audit an HTML File

content-accessibility-utility audit --input document.html --output audit_report.json --format json --severity major

Audit with the browser-backed (rendered) pass

# Add computed-style / interactive detection (requires the [rendered] extra
# and `playwright install chromium`)
content-accessibility-utility audit --input document.html --output audit_report.json --rendered

# Use the browser-backed agent to also fix and verify (requires the [agent] extra)
content-accessibility-utility audit --input document.html --output audit_report.json --agent

Remediate Accessibility Issues

content-accessibility-utility remediate --input document.html --output remediated.html --auto-fix --model-id us.anthropic.claude-sonnet-5

Translate Output (i18n)

Requires the [i18n] extra. See the CLI Guide for the full option list.

# One accessible file per language
content-accessibility-utility translate --input remediated.html --output out/ --target-languages es,fr,ja

# A single multilingual document with a language selector
content-accessibility-utility translate --input remediated.html --output multilingual.html --target-languages es,fr,ar --multilingual

Process Command (Full Pipeline)

The process command combines conversion, auditing, and remediation into a single workflow. It accepts parameters from all three operations, using the following pattern:

content-accessibility-utility process --input document.pdf --output output_dir/ [options]

Common configurations include:

# Basic process with default settings
content-accessibility-utility process --input document.pdf --output output_dir/

# Process with specific AWS settings
content-accessibility-utility process --input document.pdf --output output_dir/ --s3-bucket my-bucket --profile my-profile

# Process with audit only (no remediation)
content-accessibility-utility process --input document.pdf --output output_dir/ --skip-remediation

# Full processing with custom settings
content-accessibility-utility process --input document.pdf --output output_dir/ --severity major --auto-fix --model-id us.anthropic.claude-sonnet-5

Parameter Value Standards

Severity Levels

The standardized severity levels are:

• minor: Low-impact issues that should be fixed but don’t severely impact accessibility
• major: Medium-impact issues that significantly affect accessibility for some users
• critical: High-impact issues that make content completely inaccessible to certain users

Issue Types

Common issue types include:

• missing-alt-text: Images without alternative text
• empty-alt-text: Images with empty alt text
• generic-alt-text: Images with generic/uninformative alt text
• contrast: Insufficient color contrast
• heading-structure: Improper heading structure or nesting
• missing-labels: Form controls without labels
• missing-lang: Missing language declaration
• empty-links: Links with no text content
• table-structure: Tables without proper headers or structure
• document-structure: Issues with overall document structure and landmarks

Issue types produced only by the browser-backed rendered audit (--rendered / --agent):

• focus-not-visible: Interactive element shows no visible focus indicator (WCAG 2.4.7)
• computed-contrast-insufficient: Contrast below threshold, computed from the full CSS cascade (WCAG 1.4.3 / 1.4.11)
• missing-accessible-name: Custom widget/control with no accessible name (WCAG 4.1.2)
• missing-aria-state: ARIA role missing its required state, e.g. aria-checked (WCAG 4.1.2)
• invalid-aria-structure: ARIA role missing its required parent, e.g. tab without tablist (WCAG 4.1.2)
• focus-order-broken: Positive tabindex distorts keyboard focus order (WCAG 2.4.3)
• duplicate-id: Colliding element ids break label[for]/aria references (WCAG 4.1.1)

Rendered / agent options

Set via CLI flags or the audit options dict / configuration file:

• rendered (--rendered): run the browser-backed rendered audit in addition to the static audit.
• agent (--agent): use the Strands agent to drive the render → fix → verify loop; implies rendered.
• browser_backend: local (default) for local Playwright Chromium, or agentcore for the managed Amazon Bedrock AgentCore browser. Can also be set with the A11Y_BROWSER_BACKEND environment variable.
• agentcore_region: AWS region for the AgentCore browser (defaults to AWS_REGION / AWS_DEFAULT_REGION).
• agentcore_browser_id: optional AgentCore browser identifier (defaults to the AWS-managed aws.browser.v1).

See the Rendered & Agent Guide for details and cloud deployment.

Translation options (i18n)

Set via translate CLI flags or the translate options dict / i18n configuration section. Requires the [i18n] extra.

• target_languages (--target-languages / --languages): required comma-separated (CLI) or list (API) of BCP-47 codes, e.g. es,fr,ja.
• source_language (--source-language): source language; auto-detected from the document when omitted.
• multilingual (--multilingual): emit one combined document with an accessible language selector instead of one file per language.
• add_language_selector (--no-language-selector to disable): show the visible selector in multilingual output. Default: True.
• use_browser_language (--no-browser-language to disable): auto-select the visitor’s browser language on first load. Default: True.
• batch_size: segments per Bedrock call. Default: 20.
• model_id (--model-id): Bedrock model ID used for translation.

Python API Examples

Full Processing Pipeline

from content_accessibility_utility_on_aws.api import process_pdf_accessibility

# Process a PDF through the full pipeline
result = process_pdf_accessibility(
    pdf_path="document.pdf",
    output_dir="output/",
    conversion_options={
        "single_file": True,
        "image_format": "png"
    },
    audit_options={
        "severity_threshold": "minor",
        "detailed": True
    },
    remediation_options={
        "model_id": "us.anthropic.claude-sonnet-5",
        "auto_fix": True
    },
    perform_audit=True,
    perform_remediation=True
)

Individual Components

from content_accessibility_utility_on_aws.api import (
    convert_pdf_to_html,
    audit_html_accessibility,
    remediate_html_accessibility
)

# Convert PDF to HTML
conversion_result = convert_pdf_to_html(
    pdf_path="document.pdf",
    output_dir="output/",
    options={
        "single_file": True,
        "image_format": "png"
    }
)

# Audit HTML for accessibility issues
audit_result = audit_html_accessibility(
    html_path="output/document.html",
    options={
        "severity_threshold": "minor",
        "detailed": True
    }
)

# Remediate accessibility issues
remediation_result = remediate_html_accessibility(
    html_path="output/document.html",
    audit_report=audit_result,
    options={
        "model_id": "us.anthropic.claude-sonnet-5",
        "auto_fix": True
    }
)

Translate Output (i18n)

from content_accessibility_utility_on_aws.api import translate_html_accessibility

# Requires the [i18n] extra. Emits one accessible file per language.
translation_result = translate_html_accessibility(
    html_path="remediated.html",
    options={
        "target_languages": ["es", "fr", "ja"],
        "multilingual": False
    },
    output_path="out/"
)

See the API Integration Guide for the full options table and return shape.

Environment Variables

The tool supports configuration through the following environment variables:

Variable	Description	Required
BDA_S3_BUCKET	S3 bucket for BDA input/output (PDF conversion only)	Yes for the PDF path
BDA_PROJECT_ARN	ARN of the BDA project (PDF conversion only)	Yes for the PDF path
AWS_REGION / AWS_DEFAULT_REGION	AWS region for AWS clients	Recommended
A11Y_BROWSER_BACKEND	local (default) or agentcore — selects the rendered/agent browser backend	No

BDA_S3_BUCKET / BDA_PROJECT_ARN are only consulted for PDF conversion; the HTML/zip audit and remediation paths do not need them. Standard AWS credential environment variables (AWS_PROFILE, etc.) are honored by boto3 as usual.

Streamlit Interface Options

When using the Streamlit web interface, options are configured through the UI rather than command line parameters. The interface provides similar functionality with an intuitive web-based experience.