Can an agent act?
Enter your URL. We probe the live page the way an autonomous AI agent does and audit it across nine checks, scoring whether an agent can reach you and complete a consequential action, or silently fails at your front door.
- Agent access
- Actionability
- Operability
- Discoverability
thecraneconsultancy.com/agentic-execution-readiness
Scoring engine v1.0 · updated 29 June 2026 · deterministic probe of agent reachability and on-page action surface
Reading the live page…
You scored in badge territory. Embed this verified mark on your site; the link points back to your live result.
The mind behind the diagnostic.
A public sample of a private standard.
This scanner probes a site the way an autonomous agent does: can it even reach the page past the CDN, can it read the action in the HTML, can it operate the form. I built it because most businesses have no honest read on whether an agent can do anything on their site at all. What it surfaces in seconds, a forensic audit takes apart in full.
Led paid media for the LUSH Cosmetics global digital transformation across sixteen markets, former Head of Paid Media and Paid Media Enablement lead at senior agencies, most recently Digital Data & Analytics lead through the Everywhen brand launch across Towergate Insurance and forty other brands. A power user of HTML, CSS, JavaScript and structured data, working the machine-readability layer directly rather than briefing it out.
The brands that win 2027 are the ones an agent can act on, not just read. My work makes you operable.
A scanner finds the obvious. The audit finds the rest.
This reads what is visible in your HTML and how the edge treats agents. A forensic execution audit goes deeper: the action and offer schema an agent needs, the edge and bot rules that let agents in, and the operability work that turns a near miss into a completed task.
Speak with NicSee how AI search optimisation works, or run the AI Search Readiness scanner.
What agentic execution readiness means.
Buyers are starting to delegate tasks to autonomous AI agents: research a supplier, get a quote, make a booking, place an order. The agent only completes the task if it can reach your page, read the action in the markup, and operate the controls. Most sites have never measured whether it can.
This scanner probes your live site the way an agent does and scores it across nine checks in four dimensions. It fetches the page under real AI user-agents to see how your edge treats them, then reads the static HTML, because agents and crawlers do not run your JavaScript. It is a readiness proxy, not a guarantee: it measures the structural conditions an agent needs, not whether any specific agent will complete a purchase.
Nine checks, four dimensions.
Agent Access
Can an agent even reach you? We fetch the page under AI user-agents and compare against a browser, catching silent CDN or WAF blocks and Pay-Per-Crawl, plus your robots.txt posture for retrieval agents.
Actionability
Is the action machine-readable? Priced Offers, Services, and Action types (OrderAction, ReserveAction, potentialAction) that describe what an agent would actually do.
Operability
Can an agent work the page? Real HTML forms with named, labelled inputs, content that is present without JavaScript, and a machine-readable contact route for hand-off.
Discoverability
Can an agent find its way? Declared agent-facing interfaces and a discoverable sitemap to enumerate your actionable pages.
An action an agent cannot take is revenue it routes elsewhere.
As agents take more of the buying journey, the action surface becomes the conversion surface. A site an agent cannot reach, read or operate is one it quietly passes over for a competitor it can transact with. The work that makes you operable to an agent is becoming the work that captures agent-driven demand.
Agentic execution questions.
What is an AI agent here?
An autonomous system acting on a person's behalf, browsing and acting across the web to complete a task: comparing suppliers, requesting a quote, booking, or buying. It reads the page programmatically rather than as a human would.
How could my site be blocked without my knowing?
Many CDNs now block AI user-agents by default at the network layer, and some charge for access (HTTP 402, Pay-Per-Crawl). That can make you invisible to agents regardless of what your robots.txt says. We fetch under AI user-agents and compare to a browser to detect it.
Do you actually complete a purchase?
No. The scan is a deterministic readiness probe. It reads whether the action is reachable and machine-readable; it never submits a form, places an order, or transacts. It measures the conditions for execution, not the outcome.
Why does client-side rendering matter so much?
Agents and crawlers generally do not run your JavaScript. If the action, form or content only appears after the browser renders it, an agent sees an empty shell and cannot act. Server-rendered or pre-rendered markup is what an agent reads.
Is this the same as the AI Search scanner?
No. That one measures retrieval: can AI find and cite you. This measures execution: can an agent act on you. They are the first two of three readiness layers, retrieval then execution then representation.
Does a high score guarantee an agent will buy?
No. It measures the structural readiness you control. Whether an agent transacts also depends on the protocol it uses, the buyer's intent, and your offer. A strong score removes the structural reasons an agent fails on your site.