MCP.so
登录
A

Accessibilityai MCP Server

@groundlogic-ai-source

基本信息

分类

开发工具

传输方式

stdio

作者

groundlogic-ai-source

提交者

Evan Britto

关于 Accessibilityai MCP Server

Open-source MCP server that scans URLs for WCAG 2.1 accessibility violations, generates AI-powered code fixes, re-verifies them, and produces full VPAT 2.5 / EN 301 549 compliance reports. Bring your own Anthropic API key.

Accessibilityai MCP Server 配置

使用下面的配置,将此服务器添加到你的 MCP 客户端。

{
  "mcpServers": {
    "accessibility-ai": {
      "url": "https://accessibility.groundlogic.ai/mcp",
      "transport": "http"
    }
  }
}

Accessibilityai MCP Server 工具

4

Scans a URL for WCAG 2.1 accessibility violations using DOM analysis (axe-core) and visual AI (Claude Vision). Returns structured violation data and a scan_id for use with other tools.

Takes scan results and generates specific code fixes for each accessibility violation. Returns ready-to-apply code patches and a copy-paste summary for Claude Code or Replit Agent.

Re-scans a URL after fixes have been applied and compares results against the original scan. Returns which violations were resolved, which persist, and any new issues introduced.

Generates a complete VPAT 2.5 EN 301 549 accessibility conformance report based on scan results. Returns structured JSON and a downloadable PDF (base64-encoded). Covers all EN 301 549 clauses; non-web clauses are marked Not Evaluated with professional notes.

概览

What is AccessibilityAI?

AccessibilityAI is an MCP server that scans any URL for WCAG 2.1 violations, generates copy-paste code fixes, re-verifies results, and produces VPAT 2.5 / EN 301 549 compliance reports. It requires your own Anthropic API key and is intended for use with AI assistants like Claude.

How to use AccessibilityAI?

Configure the server URL (https://accessibility.groundlogic.ai/mcp) with HTTP transport in your MCP client. Then send prompts like "Scan https://mysite.com for accessibility issues and generate fixes" to invoke the four tools: scan_accessibility, generate_fixes, re_verify, and generate_vpat.

Key features of AccessibilityAI

  • Scans any URL for WCAG 2.1 violations using axe-core and Claude Vision.
  • Generates AI-powered copy-paste code fixes for each violation.
  • Re-verifies after fixes, showing resolved, persisting, and new violations.
  • Produces complete VPAT 2.5 / EN 301 549 reports as JSON and PDF.
  • Requires your own Anthropic API key; key never stored.
  • Batches fix generation 5 per call for reliability.

Use cases of AccessibilityAI

  • Scan a website for accessibility compliance before launch.
  • Generate copy-paste code fixes to address WCAG violations.
  • Re-verify a site after applying fixes to confirm resolution.
  • Produce a VPAT report for enterprise procurement or auditors.
  • Automate accessibility testing and reporting within an AI assistant.

常见问题

1

How do I connect to AccessibilityAI and what authentication is needed?

The server uses HTTP transport at `https://accessibility.groundlogic.ai/mcp`. You must provide your own Anthropic API key (BYOK); the key is never stored.

2

What tools does AccessibilityAI provide?

It provides four MCP tools: `scan_accessibility` (scan a URL), `generate_fixes` (generate code fixes for a scan), `re_verify` (rescan after fixes), and `generate_vpat` (produce a VPAT report).

3

Does AccessibilityAI store my data or API key?

No, your Anthropic API key is never stored. Scan results and fix data are handled per request.

4

What format does the VPAT report come in?

The VPAT 2.5 / EN 301 549 report is returned as both JSON and PDF.

5

How are the code fixes generated and delivered?

Fixes are AI-generated based on the scan results and are batched 5 per call for reliability. They are intended as copy-paste code snippets.

评论

开发工具 分类下的更多 MCP 服务器