Obenan Briefing · Signal
The web is being scored for agent readiness. For a local business, the score still rests on merchant truth.
On April 17, 2026, Cloudflare launched an Agent Readiness score and a Radar dataset that turn agent-facing web standards into something operators can measure. Its scan of 200,000 domains showed how rare the new signals still are. The score grades whether a site can be discovered, fetched, and read by agents. It does not grade whether the local business behind the page tells the truth.
Published May 29, 2026
The one line
A readiness score measures whether an agent can reach and read your site. It cannot measure whether what your site says about hours, availability, and policy is actually true. For a local business, the second half is the half that decides the answer.
- Published
- May 29, 2026
- Format
- Signal briefing
- Sources
- 5 primary, public
- Coverage
- April 17 to 24, 2026
Public sources only. Observed facts and our interpretation are labeled separately.
The 60-second read
Three questions, three answers.
What changed at the edge, why a measured readiness standard matters to anyone making local businesses legible to agents, and the part the score cannot reach.
Agent readiness became a measurable, public standard.
On April 17, 2026, Cloudflare launched an Agent Readiness score and a Radar dataset that benchmark agent-facing web standards across the open web: discoverability, machine-readable delivery, bot-access control, and protocol discovery. Live checks on April 24, 2026 confirmed the surface is exposed through public widgets and machine-readable endpoints, not only a blog post.
Legibility for agents stopped being a belief and became a benchmark.
Cloudflare reported that its scan of 200,000 domains found robots.txt on 78 percent of sites, Content Signals on 4 percent, Markdown negotiation on 3.9 percent, and MCP Server Cards plus API Catalogs on fewer than 15 sites in the entire scan. The standards are real and measurable, and adoption is still early.
Whether the truth behind the page is correct, complete, and current.
A high readiness score proves an agent can find and fetch your site. It says nothing about whether the listed hours are right, the slot is open, or the policy will hold. For a local business, that is the part an agent actually repeats to a customer.
What to do with this
Five moves for the people making this layer ready.
Split by what a merchant controls and what a platform controls. The two are not the same job, and confusing them is how a clean score still produces a wrong answer.
- 01
Readiness is necessary, not sufficient (merchant-controlled)
Make the site discoverable, fetchable, and machine-readable. It is table stakes now that the standard is measured. But a clean score is the floor, not proof that the answer about your business will be right.
- 02
The facts an agent repeats are merchant-authored (merchant-controlled)
Hours, address, services, availability, and policy are yours to make true. No edge standard supplies them. The edge can certify the delivery; only the merchant can certify the content.
- 03
Edge standards are platform-controlled, not merchant truth (platform-controlled)
robots.txt, Content Signals, markdown negotiation, and bot policy govern access and delivery. They route and gate the request. They do not attest that the business on the other side is described correctly.
- 04
A score is a diagnostic, not a ranking lever (shared)
Treat the Agent Readiness score as a checklist for legibility, not as a promise of citation. Passing it does not move what ChatGPT, Gemini, Perplexity, or Google Maps decide to recommend.
- 05
Separate readiness from recommendation when you report (shared)
Keep two columns. Did the site become more legible to agents, and did the answer about the business get more accurate. Conflating them is how teams over-claim progress.
The evidence, dated
One day at the edge, four moving parts.
One launch, four public artifacts, shown in the order Cloudflare published them. Read together, the shift from belief to benchmark is hard to miss.
An Agent Readiness score launches
Cloudflare launched an Agent Readiness score and isitagentready.com, bundling four dimensions: discoverability through robots.txt, sitemaps, and link headers; machine-readable delivery through Accept text/markdown; bot-access control through Content Signals, AI bot rules, and Web Bot Auth; and protocol discovery through API catalogs, OAuth discovery, MCP server cards, and WebMCP.
A 200,000-domain benchmark
Cloudflare published a scan of 200,000 domains. It reported robots.txt on 78 percent of sites, Content Signals on 4 percent, Markdown negotiation on 3.9 percent, and MCP Server Cards plus API Catalogs on fewer than 15 sites across the whole scan. The distance between basic and agent-native standards is wide.
Crawl control gains diagnostics
On the same day, Cloudflare expanded AI Crawl Control. Redirects for AI Training turns canonical tags into 301 redirects for verified training crawlers, a Content Format chart compares requested versus served content types, and the renamed Directives tab links to an Agent Readiness check. Control and measurement both moved at the edge.
It is exposed, not just announced
On April 24, 2026, a markdown-negotiation request returned content-type text/markdown with token and content-signal headers, and isitagentready.com exposed public MCP metadata and a machine-readable skills index covering scan-site, robots-txt, content-signals, sitemap, llms-txt, and markdown-negotiation. The readiness surface is live product behavior, not a concept.
In one day, agent readiness went from a publishing belief to a measured standard with a public scorecard, an edge control layer, and live machine-readable endpoints. What the standard measures is delivery and discovery. What it leaves to the business is whether the delivered facts are true.
The distinction that matters
A readiness score is not a true answer.
A high score proves an agent can discover your site, fetch it efficiently, read it as machine-readable content, and understand your bot policy. The Cloudflare launch made that measurable. A true answer is something else: that the specific fact an agent repeats about your business, the hours, the slot, the price, the policy, is correct, complete, and current at the moment it is asked.
For a static brochure page the gap is small. For a local business whose hours change, whose calendar fills, and whose policy has exceptions, the gap is the whole problem. Merchant truth is its own layer, and it is the one piece no readiness score, crawl-control setting, or markdown endpoint supplies.
Accurate identity
Name, address, and category that match reality across the surfaces an agent reads.
Current hours and availability
Open hours, special hours, and real availability, not a listing that drifted out of date.
Complete policy
Booking, deposit, cancellation, and service terms an agent can act on without guessing.
Services and attributes
What the business actually offers, described the way a customer would ask for it.
Freshness
Facts that are current at the moment of the answer, not last quarter's snapshot.
Evidence
Reviews and signals that corroborate the claim, so the answer an agent gives can be trusted.
Where it breaks today
When the score is green and the answer is wrong.
Each of these can happen on a site that passes an agent-readiness check. The delivery worked. The fact behind it did not.
An agent reads perfectly machine-readable hours that are three weeks out of date, and sends a customer to a closed door.
An agent fetches a clean, markdown-negotiated page that lists a service the business stopped offering.
An agent honors a bot policy and a sitemap flawlessly, then quotes a cancellation rule the merchant will not stand behind.
An agent confirms a slot from a fast, well-structured page while the merchant's own calendar shows it already booked.
In every case the site was agent-ready by the standard. The answer was still wrong, because readiness measured the delivery, not the truth. That gap is merchant-authored, and a green score does not close it.
Where the work sits
The readiness stack, and the layer in the middle.
Agent readiness has three layers. The outer two are being standardized and measured at the edge by platforms such as Cloudflare. The middle layer is merchant-authored, and it does not arrive with a score.
Discovery and delivery decide whether an agent can reach and read the page. The answer engine decides whether to cite or recommend. Between them sits the question everything depends on: is what the page says about this business true, complete, and current.
Agents can reach and read the site
robots.txt, sitemaps, markdown negotiation, bot policy, and protocol discovery. Now scored and benchmarked at the edge.
Actor: edge platforms
Is the published fact true
Accurate identity, current hours, real availability, complete policy, and corroborating evidence. This is where Obenan works.
Actor: the merchant
The agent cites or recommends
ChatGPT, Gemini, Perplexity, and Google Maps decide what to surface and recommend. Platform-controlled, and not bought with a score.
Actor: answer engines
Discovery and delivery are becoming measured infrastructure. The answer is platform-controlled. Merchant truth, the layer between them, is still merchant-authored work.
Obenan's view
An edge can measure whether an agent can read your site. Only a merchant can make what it reads true.
The agent-readiness story is being told as a standards-and-scoring story. It is really a trust story, and the trust begins inside the merchant, before any agent reads the page. Cloudflare's launch is real and it matters. It makes legibility measurable and gives operators a useful checklist. It does not make the facts on the page true, and it does not decide what an answer engine recommends.
Obenan works on that middle layer: the accuracy, completeness, freshness, and committability of what a local business publishes about itself, and the reviews and signals that corroborate it. We do not run the edge and we do not rank the answer. We make the facts a business stands behind correct and machine-legible, so that when an agent reads a ready site, the answer it gives is one the business can actually keep.
How to read this
Observed, inferred, and watched. Kept separate on purpose.
This is a Signal briefing. We report what the public record shows, we label where we are interpreting, and we name what would change our read.
Observed
Cloudflare's April 17, 2026 launch describes the Agent Readiness score, the four readiness dimensions, the 200,000-domain scan with its adoption percentages, and the AI Crawl Control diagnostics, on Cloudflare's own public blog, changelog, and docs. Live checks on April 24, 2026 returned markdown content-type and content-signal headers and exposed public MCP metadata.
Inferred
That a readiness score measures delivery and discovery but not the accuracy of merchant facts, and that merchant truth therefore remains the deciding layer for local businesses, is Obenan's reading of the public evidence, not a statement by Cloudflare.
What we are watching
Whether any agent-readiness standard begins to score the accuracy or freshness of published business facts, not just their delivery. The first that does narrows the gap this briefing describes.
What this briefing does not claim
The boundaries, stated plainly.
Precision protects the reader and the companies named. These are the lines this briefing does not cross.
- 01
A high Cloudflare Agent Readiness score does not prove that ChatGPT, Gemini, Perplexity, or Google Maps will cite or recommend a business. It measures legibility, not recommendation.
- 02
Agent-readiness standards measure discovery and delivery. They do not validate that a site's hours, availability, services, or policy are accurate.
- 03
Obenan is not affiliated with, partnered with, or endorsed by Cloudflare, and is not part of the Agent Readiness program or isitagentready.com.
- 04
Obenan does not operate edge infrastructure, crawl control, or content delivery, and does not control what any answer engine ranks or recommends.
- 05
Nothing here implies approval, endorsement, partnership, pilot, certification, or integration with any company named in this briefing.
Keep reading
The thesis and the action.
If you are making local businesses ready for agents, the score is the easy half.
We work on the merchant-truth layer a readiness score cannot measure: accuracy, freshness, availability, policy, and the evidence behind them. If that is your problem too, let's compare notes.
Public research. No partner status is claimed or implied.
Sources
All primary and public. Each can be inspected directly. Developed dates are the public publication or update dates. Checked dates are April 24, 2026.
Primary sources
- 1.Cloudflare: Agent Readinessblog.cloudflare.com · April 17, 2026 · checked April 24, 2026
Primary. Cloudflare's launch of the Agent Readiness score, the four readiness dimensions, and the 200,000-domain benchmark.
- 2.Cloudflare: Redirects for AI Training (changelog)developers.cloudflare.com · April 17, 2026 · checked April 17, 2026
Primary. AI Crawl Control adds Redirects for AI Training, a Content Format chart, and a Directives tab linking an Agent Readiness check.
- 3.Cloudflare: is it agent readyisitagentready.com · April 17, 2026 · checked April 24, 2026
Primary. The public Agent Readiness checker launched alongside the score, exposing machine-readable MCP metadata and a skills index.
- 4.Cloudflare AI Crawl Control documentationdevelopers.cloudflare.com · April 20, 2026 · checked April 24, 2026
Primary, product reference. Crawl-control directives, content-format diagnostics, and the link to an Agent Readiness check.
- 5.Cloudflare Radarradar.cloudflare.com · April 17, 2026 · checked April 24, 2026
Primary, dataset reference. Radar AI Insights now includes agent-standards adoption and AI bot response-status distributions.
Obenan Briefings. AI Visibility lane. This is a Signal briefing: a dated, bounded market update with a named operator implication. It is not investment advice and implies no relationship with any company named.