axle · the honest case for paying

If axe-core is free, why pay for axle?

Fair question. The comparison pagetells you Pa11y, Lighthouse CI, and the axe-core CLI all run the same engine for $0. The free tier of axle runs the same engine too. Here's when paying — for axle or any other paid tool — actually pays back.

The four reasons people actually upgrade

1. Claude-generated source-code fix PRs

Every scanner gives you a list of violations. color-contrast on 17 elements, image-alt missing on 23 images, label not associated with input. Great. Now someone has to fix all of that.

Average dev time per fix in our internal benchmark: 22 minutes. At 47 violations and $100/hr loaded engineering rate, that's $1,720 of engineering work per scan. The first scan on most apps surfaces 30-120 violations. The math gets ugly fast.

axle Team ($49/mo) ships hosted Claude fix suggestions on every PR comment, plus the create-issues mode that opens / updates / closes a GitHub Issue per violation with the AI-suggested diff in the body. Teams triage in their existing board (Linear / Jira sync via GitHub mirroring) instead of in ephemeral PR threads.

ROI calc:Average remediation time per violation is ~22 minutes. A 47-violation scan is ~17 hours of engineering at $100/hr = $1,700. Even if Claude's suggestion saves 30% of that, it's ~$500 in eng cost recovered per scan — and the team triages in their actual issue board instead of from a one-off PR comment. The Team plan pays back in month one.

Honest scope: auto-opening real fix PRs (not just Issues) — mapping the AI diff back to the source file — is in development. See /roadmap for status.

2. EU multilingual statement pack (EAA 2025) partial

If you ship a consumer-facing service into the EU and you employ 10+ people, EAA 2025 has been enforceable since 28 June 2025. Penalties are state-by-state: DE €100K, FR €25K, IT 5% of turnover, ES €600K, NL €900K, PL 10% of turnover.

Each member state requires the accessibility statement in its own language, with the local conformance authority, escalation procedure, and named contact. One English statement doesn't satisfy the rules.

axle Business ($299/mo) is being built out as a statement pack covering German, French, Italian, Spanish, Dutch, Portuguese, Swedish, Polish, and Hebrew (Israeli תקנה 35) with per-country authority + escalation pre-filled. Hebrew ships today; the EU languages are in the rollout pipeline — see /roadmap.

ROI calc:One legal-team hour to draft a single jurisdiction's statement = $400+. A Business-tier customer ships into 6+ EU markets typically. axle Business covers an entire year of regenerated multi-jurisdiction statements for $3,588 — typically less than what one accessibility lawyer charges to review a single statement.

3. Auto-managed GitHub Issues per violation

Most scanners stop at a PR comment that disappears once the PR merges. With create-issues: true, axle opens a deduplicated GitHub Issue per WCAG violation — impact-labelled, AI-fix in the body, target selector + first failing element preserved. When the next scan no longer sees the violation, axle auto-closes the issue with a comment linking to the fixing commit.

Net effect: your team triages accessibility in the same board as everything else (Linear / Jira / Notion sync via GitHub's native mirroring) instead of in an ephemeral PR thread. Treat it like Renovate but for WCAG.

ROI calc: One agency we talked to was paying a contractor 4 hr/month to copy violations from PR comments into their Linear board. axle Team ($49/mo) replaces that workflow entirely — about $400 of manual ops eliminated per month per repo.

4. MCP server for Claude Desktop / Cursor / Cline

New surface, shipped May 2026. axle-mcp is the only Model Context Protocol server for WCAG scanning. Wire it into any MCP-compatible agent in two lines:

{
  "mcpServers": {
    "axle": {
      "command": "npx",
      "args": ["-y", "axle-mcp"]
    }
  }
}

Now Claude can scan a URL during code review, explain a violation with WCAG mapping, and propose remediation patterns — all in chat, no context-switching to a CLI. For teams using Cursor / Claude Desktop as their primary surface, this is the only way to get accessibility into the loop without breaking flow.

ROI calc:Hard to put a number on agent ergonomics. But every team we've shown this to has installed it within 30 seconds of seeing the demo. Try the free tier (5 scans/day, no API key) before paying.

When NOT to pay for axle

• You ship to one repo, US-only, and don't need AI-generated fixes. The free tier covers you. Use it forever. No upsell pressure, no degradation.
• You already use axe DevTools Pro and the team budget is there. Deque's product is mature; axle isn't trying to replace it for established enterprise customers. Run both in CI for a month, see what each catches.
• You want a managed dashboard with multi-user RBAC, SAML, audit logs. That's not what we built. Look at Tenon.io or axe Watcher.
• You believe an overlay widget fixes accessibility. It doesn't. Here's the FTC settlement.

The honest social-proof signal

axle is 30 days old. As of writing: 12 distribution surfaces shipped (GitHub Action, npm CLI, 3 hosting plugins, WordPress plugin, VSCode extension, Storybook addon, MCP server, ChatGPT GPT, integrations recipes). 0 paid customers yet. We're telling you that because if you're looking for “1,000 teams trust us,” we don't have that signal yet.

What we do have: the FTC settled against accessiBe for $1M in January 2025; EAA 2025 enforcement kicked in June 2025; ADA Title III lawsuits hit 4,000+ federal filings in 2024. The tooling that exists for handling these regulatory pressures is either expensive ($5-20K/year overlays that don't work; $40K+ audit firms) or free-and-DIY (Pa11y, axe-core CLI, no support, no statements, no AI). axle is the middle tier for teams that need to ship and not get sued.