Developer & AI interfaces

Accessibility, seamlessly integrated into your development workflow

Automated WCAG validation directly in the IDE, anchored for AI coding agents and in the CI/CD pipeline. Use our open‑source local MCP for free scans on localhost, query live audit data in your editor and secure every release automatically.

Start testing for free

Model Context Protocol (MCP)

Connect accessibility directly to your AI agents

navable provides two dedicated MCP servers: validate local code on localhost during development or query live audit data without leaving your editor.

Open source

Local MCP – validate localhost in the dev workflow

Gives your AI coding agents deep real‑browser scans via Playwright and axe‑core (optionally Pa11y/HTMLCS). The server returns WCAG 2.1 A+AA violations including automatic EN 301 549 mapping and a structured fix plan.

Tools: run_accessibility_scan, generate_fix_plan, update_fix_status
Generates a .navable-plan.json that your AI agent can execute autonomously

claude mcp add navable -- npx -y @navable/mcp

Pro plan

Hosted MCP – global audit data directly in your IDE

Connect VS Code, Cursor or Windsurf to your central production and staging audits. Ask your AI for current scores, top issues and trends across the organisation – without changing context.

Query monitored URL collections, scores and severities directly
Secure endpoint via mcp.navable.io with x‑api‑key authentication

https://mcp.navable.io/   (Header: x-api-key)

Set up Local MCP and check its test limits

Open-source agent skills

A repeatable scan → fix → verify workflow

Pair the MCP with @navable/skills so your agent follows a consistent process instead of improvising — with detailed fix guides for ARIA, color, forms, headings, images, keyboard, landmarks, language, navigation and tables.

scan-accessibility

Scan a URL, generate a fix plan, apply fixes and verify.

fix-accessibility

Resume an existing .navable-plan.json and work through pending items.

review-component

Review a component's source for accessibility issues — no browser needed.

audit-page-structure

Audit landmarks, heading hierarchy, navigation and document metadata.

Install into your agent's skills folder

cp -r skills/* .github/skills/   # Copilot
cp -r skills/* .claude/skills/   # Claude Code
cp -r skills/* .agents/skills/   # Cursor

CI/CD & DevOps automation

Secure every deployment for accessibility – fully automated

Integrate WCAG checks directly into your existing deployment pipeline. Trigger audits programmatically, use automated quality gates before release, or forward real‑time results via webhooks.

Automated triggers: run deep real‑browser audits after every deploy
Dynamic URL management: add new paths directly from your code
Real‑time interfaces with secure header authentication
Webhook pipelines: send audit events straight to your internal tools

curl -X POST https://app.navable.io/api/... \
  -H "x-api-key: $navable_API_KEY"

IDE fix-prompts

Every finding ships with a fix prompt

Paste a ready-made prompt into Copilot, Cursor or ChatGPT. The issue, affected code and fix instructions are included. No accessibility expertise required.

Click to see a real example navable generates.

Built-in accessibility knowledge

Your agent gets expert context, not guesses

The open-source MCP exposes curated, WCAG-grounded resources so agents fix issues correctly the first time.

WCAG → EN 301 549 mapping

Success criteria mapped to EN 301 549, testability and axe rules.

>50 fix patterns

Per axe-rule before/after code patterns your agent can apply.

WAI-ARIA APG patterns

Authoring-practices guidance for dialogs, comboboxes and more.

Semantic HTML guides

Per-element implicit roles and correct usage.

BFSG legal context

Scope, glossary and enforcement context for German compliance.

Bring accessibility into your stack

Start free with the open-source MCP, or unlock hosted audit data and the API on Pro.

Frequently Asked Questions

Yes. The local MCP setup (@navable/mcp) and the agent skills (@navable/skills) are open-source. You can scan localhost and run the fix workflow without an account. The hosted MCP (live audit data) and the REST API are part of the Pro plan.

Any MCP-compatible client — including VS Code + GitHub Copilot, Cursor, Windsurf and Claude Code. Skills install into .github/skills (Copilot), .claude/skills (Claude Code) or .agents/skills (Cursor).

It runs a real browser with axe-core against your localhost pages and returns WCAG 2.1 Level A + AA violations with EN 301 549 mapping. You can optionally enable a second engine (Pa11y/HTMLCS) for higher-confidence, deduplicated findings.

Use the REST API to trigger an audit after each deploy and fail the build on critical issues. Business plans add webhook integration to push results into your tools.

No. The local MCP scans your running dev server in your own environment and writes the fix plan to your repo. Nothing is uploaded for the open-source scanning workflow.