Add agent-ready.dev to Websites

[mlava]

Adds agent-ready.dev to the list.

Why it fits: agent-ready.dev is a live site that implements WebMCP — it registers its own tools (scan_site, get_scan, ask) on navigator.modelContext (feature-detected, no-op where unsupported), so in-browser agents can scan any public URL for AI-agent readability and get back a 0–100 score with per-check fixes. It scores sites against the Vercel Agent Readability Spec, llmstxt.org, and agent-protocol manifests.

• Single entry, follows the existing format in the section.
• Public, free to use, no signup required to scan.

🤖 Generated with Claude Code

[mlava]

Gentle bump — still cleanly mergeable, no conflicts.

Adding the thing I should have led with. The distinction that matters for a WebMCP list isn't that the site has an audit feature; it's what the registered tool hands back to the agent. scan_site / get_scan return the full structured scan — score, per-check status, remediation — straight through navigator.modelContext, on every call, with no signup and no rate gate. The agent gets data it can act on, not a string telling the human to go look at a UI. That round-trip is the whole reason the surface is worth listing.

Third tool is ask: natural-language search over the scoring methodology, the check registry, and the specs we validate against, so an agent can look up why a check failed without leaving the tool surface.

On fit alongside isainative.dev — they're complementary rather than overlapping: that one audits a public GitHub repo for AI-coding readiness, this one audits a live site for agent readability (Vercel Agent Readability Spec, llmstxt.org, and agent-protocol manifests — MCP server cards, A2A, agents.json, agent-permissions.json).

For whatever it's worth as a signal, webfuse-com/awesome-webmcp merged the equivalent entry last week on exactly this reasoning (#6).

Happy to reword, shorten, or move it to a different section if Websites isn't the right home.