Point your agent at the guide
Use the folder with Claude Code, Codex, Cursor, OpenCode, Aider, or another agent that can read local instructions.
What you are looking at
One Next.js demo app, taken from 0 to 100.
The framework did not just add a badge. It added the public files, headers, well-known endpoints, browser tools, and verification loop that agents and scanners can actually inspect. The same pattern works as a report-only audit or as an approved code patch for your own website.
Live AI-Readiness Score
Scanning this site against isitagentready.com...
Make Your Website Agent-Ready
Ask for report or patch
Choose a readiness report, an approved patch run, or both. The guide asks before changing files or hosted settings.
Deploy and verify
Ship the next tier, scan the production URL, then keep going until the evidence improves.
Read improve-ai-readiness/SKILL.md for https://your-site.com. First generate a readiness report, then ask before applying fixes.The public audit found no readiness surface: no robots, sitemap, markdown, or well-known discovery.
The same deployed app now reaches Level 5 with fourteen passing checks and commerce left neutral.
Use the framework to get a report, approve a patch plan, deploy, and re-scan.
What This Demonstrates
Read the field guideLevel 1i
Crawl contract
robots.txt, sitemap.xml, and Link headers tell crawlers where the site map and machine endpoints live.
Level 3i
Readable pages
Markdown negotiation gives agents a clean representation without asking them to parse the visual page.
Level 4i
Agent directory
MCP, A2A, agent-skills, and api-catalog endpoints publish the actions and documentation agents can use.
Level 5i
Native controls
OAuth metadata, Web Bot Auth, and browser WebMCP move the site from static discovery into governed interaction.
Why The Score Improves
More detailMore crawl certainty
Robots rules, sitemap entries, canonical links, and Link headers reduce guesswork for crawlers and scanners.
Cleaner agent reading
Markdown negotiation and guidance files give agents a stable text path instead of forcing them through visual HTML.
Higher action discovery
Well-known JSON, Agent Skills, MCP, A2A, OpenAPI, and WebMCP expose what the site can do, where it lives, and how to call it.
WebMCP Browser Tools
checking...Tools are registered on page load through navigator.modelContext.registerTool(). The fallback polyfill also supports Chromium builds that expose the test harness.
Demo Well-Known Endpoints
live-checked against this origin| Path | Tier | Status | Purpose |
|---|---|---|---|
/robots.txt | L1 | checking... | Crawler allow/disallow rules plus sitemap pointer |
/sitemap.xml | L1 | checking... | Machine-readable URL inventory |
/llms.txt | L2 | checking... | LLM-friendly site summary |
/AGENTS.md | L2 | checking... | Agent-readable site guide |
/.well-known/mcp/server-card.json | L3 | checking... | MCP server card for model context tools |
/.well-known/mcp.json | L3 | checking... | Fallback MCP server card path accepted by the audit |
/.well-known/agent-card.json | L4 | checking... | A2A agent card for agent-to-agent discovery |
/.well-known/agent-skills/index.json | L4 | checking... | Skills discovery index |
/.well-known/api-catalog | L5 | checking... | API catalog |
/.well-known/oauth-protected-resource | L5 | checking... | OAuth Protected Resource Metadata |
/.well-known/oauth-authorization-server | L5 | checking... | OAuth Authorization Server Metadata |
/.well-known/http-message-signatures-directory | L5 | checking... | Web Bot Auth public key directory |
Score API
cached 5 min, same originGET /api/score proxies the live audit for this demo.GET /api/score?url=https://example.com scans another public URL.
Use Case
runbook, not a checklist dumpThe framework audits first, ships only the next tier of files, waits for deployment, and scans again. It can stop at a report, or it can apply a patch after the user approves the plan.
Beyond One Checker
iThe same readiness surface helps with Cloudflare's audit, SiteDex-style checks, agent directories, SEO crawlers, citation systems, and future AIPREF-style content usage policies. llms.txt is included because it is cheap, but the durable work is crawl policy, structured data, markdown access, signed bot policy, and tool discovery.