How xebia/accessibilityanalyzer works
Project type and purpose
• This is a full-stack, service-oriented application (no UI) consisting of a Web API plus an AI-powered analysis engine.
• Purpose: accept a target URL or HTML payload, fetch/parse its content, run WCAG accessibility checks (HTML structure + visual analysis), and return a structured JSON report.
File structure
• Top‐level directories:
• AccessibilityAnalyzer/ ← ASP .NET Core Web API project
• AccessibilityAnalyzer.Ai/ ← Core analysis library (process orchestration + step implementations)
• compose.yaml ← Docker Compose orchestration for API + OmniParser server
• AccessibilityAnalyzer (API) contains
• Program.cs, Analyzer.cs ← API entry-point + controller/service definitions
• Extensions/ServicesExtensions ← DI registration (kernel, HTTP clients, pipelines)
• Dto/Analysis.cs ← request/response data contracts
• appsettings*.json, .env ← configuration (API keys, endpoints)
• Dockerfile ← containerization of the API
• AccessibilityAnalyzer.Ai contains
• OmniParserLocalService.cs ← HTTP client wrapper for the external OmniParser server
• Processes/*.cs ← top-level orchestrators (HtmlAnalysisProcess, VisualAnalysisProcess)
• Steps/… ← individual “analysis steps” (chain of responsibility)
• AnalysisAggregationStep.cs ← merges results from all steps
• External: OmniParser server (Python + Playwright) is spun up separately (referenced in README & compose.yaml).
README for additional insights
• Leverages Microsoft Semantic Kernel (SK) for LLM orchestrations.
• Playwright (via OmniParser) for page rendering & DOM snapshots.
• Produces aggregated JSON with detailed issue listings.