AVR v1.1.0 Full 13-Section Audit: Cloudflare Radar Data Meets AI Visibility

What does a 13-section AI visibility audit actually look like?

AVR v1.1.0 expanded from 8 sections to 13 by adding 5 checks grounded in Cloudflare Radar data from May 2026. These are not theoretical checks. Each one measures your site against an empirical baseline drawn from traffic patterns across Cloudflare’s network. I ran the complete audit against chudi.dev to see what a real site looks like against these baselines.

What are the 5 new Cloudflare-derived checks?

Cloudflare Radar publishes weekly AI bot traffic data. The May 17-23, 2026 window shows that AI crawlers now represent a significant share of all bot traffic (Googlebot 26.2%, GPTBot 10.5%, ClaudeBot 9.3%), and that 8.7% of AI bot requests get blocked with 403 Forbidden. The 5 new AVR checks test whether your site falls in that blocked percentage or not.

Bot Response Code (§2.8) sends GET requests with ClaudeBot, GPTBot, Bingbot, and PerplexityBot user-agent strings and checks whether each gets a 200 OK or a 403/429 block. chudi.dev result: ACCESS-OPEN, 4/4 bots get 200.

Markdown Negotiation (§2.9) tests whether your server responds to Accept: text/markdown with actual markdown content. Cloudflare data shows only 5.3% of top sites support this, and the payload reduction is 93% (down to 7% of HTML size). chudi.dev result: MARKDOWN-READY. The server returns a curated markdown summary with recent posts, topics, and author info.

AI Rules in robots.txt (§2.10) parses your robots.txt for AI-specific user-agent directives. 79% of top domains have them, but many are incomplete (blocking GPTBot but not ClaudeBot, or vice versa). chudi.dev result: AI-RULES-COMPLETE, 3/3 checks pass. All major AI crawlers have explicit directives.

Agent Readiness Tier (§2.11) scores your site 0-4 across four agent infrastructure standards: robots.txt AI rules, XML sitemap, WebMCP manifest, and A2A AgentCard. WebMCP has 0% adoption among the top 200K domains scanned. chudi.dev result: AGENT-TIER-HIGH, 4/4. All four standards are implemented.

Crawl Signal (§2.12) tests response time under a ClaudeBot user-agent (under 3 seconds threshold), confirms robots.txt does not fully block major AI crawlers, and checks for aggressive rate limiting. chudi.dev result: CRAWL-ACCESSIBLE, 3/3 checks pass.

What did the full 13-section audit show?

Here is the complete chudi.dev audit from May 23, 2026:

Section	Verdict	Score
§1 SEO Foundation	FAIL	Lighthouse skipped
§2 AI Infrastructure	PARTIAL	Needs work
§2.7 Agent Readiness (WebMCP)	AGENT-READY	3/3
§3 Fact-Block Density	EXTRACTABLE	100/100
§7 Citation Decay Rate	GROWING	HIGH confidence
§2.8 Bot Response Code	ACCESS-OPEN	4/4
§2.9 Markdown Negotiation	MARKDOWN-READY	1/2
§2.10 AI Rules in robots.txt	AI-RULES-COMPLETE	3/3
§2.11 Agent Readiness Tier	AGENT-TIER-HIGH	4/4
§2.12 Crawl Signal	CRAWL-ACCESSIBLE	3/3

The Cloudflare-derived checks (§2.8-§2.12) all pass. The site is ahead of the market on every agent standard that Cloudflare measures.

Fact-Block Density is 100/100 EXTRACTABLE. The root page passed all five checks (standalone answer, first-200-tokens, 40-60 word band, question-format headings, FAQ section) after the prior remediation arc converted H2s to questions and added FAQ sections.

Why does the Citation Decay Rate matter?

Citation Decay Rate is a first-to-market metric in AVR v1.1.0. No Semrush, BrightEdge, or Conductor product tracks it. The metric measures whether your citations are holding, growing, or decaying over time. chudi.dev shows GROWING with HIGH confidence, meaning AI engines are citing the site more frequently over time, not less.

This aligns with Cloudflare’s crawl-to-refer data: ClaudeBot’s ratio is 10,600:1 but declining 16.9%, meaning Claude is sending more referral traffic per crawl. The product thesis behind citability.dev is that making your content more citable leads to more AI-driven referrals. The Citation Decay Rate metric proves whether that is actually happening for a given site.

What does the citability.dev free scan show now?

The free scan at citability.dev now runs 15 checks (up from 10). The 5 new Cloudflare-backed checks are:

1. AI Bot Access - are ClaudeBot and GPTBot blocked?
2. Markdown Negotiation - does your server support Accept: text/markdown?
3. AI Rules in robots.txt - do you have AI-specific directives?
4. Agent Readiness Tier - robots.txt + sitemap + WebMCP + AgentCard (0-4 score)
5. Crawl Responsiveness - is your site fast enough for AI crawlers?

Every check runs against Cloudflare Radar baselines. When the scan says “8.7% of AI bot requests get 403 globally,” that is an empirical number from Cloudflare’s network, not a made-up benchmark.

What is next for AVR v1.1.0?

The framework now has 13 audit sections covering everything from basic SEO infrastructure through agent readiness to citation decay tracking. The remaining work is remediation: fixing the Fact-Block Density regression on the SSR root page, expanding markdown negotiation to blog post routes, and building the Citation Decay Rate into citability.dev’s paid reporting tiers.

The audit scripts, the Cloudflare empirics glossary, and all 7 codex nodes are open source at github.com/ChudiNnorukam/avr-pipeline.