---
title: "If your site is not agent ready | Agent Readiness Compare"
description: "Ten tasks an AI agent will not complete on a site that is not agent ready, from getting past robots.txt to paying, with the fix for each and how to check it."
canonical: https://agentreadinesscompare.com/10-things-if-your-site-is-not-agent-ready/
site: "Agent Readiness Compare"
last_updated: 2026-11-05
section: "Reference"
---

# If your site is not agent ready | Agent Readiness Compare

> Ten tasks an AI agent will not complete on a site that is not agent ready, from getting past robots.txt to paying, with the fix for each and how to check it.

Source: https://agentreadinesscompare.com/10-things-if-your-site-is-not-agent-ready/
Site: Agent Readiness Compare
Last reviewed: 2026-11-05

# 10 things that won't happen if your site isn't agent ready

Where an agent stops, and what to fix so it does not.

In short

An agent sent to your site tries to get in, find the right pages, read your pricing, compare plans, search your docs, fill in a form, sign up, call an action, prove who it is and pay. Each step can stop it. Below are ten of those stops, the fix for each and how to check it.

## 1. How much evidence stands behind the ten

Failure modes, not ratesTwo of these stops have measured evidence behind them: answers hidden in JavaScript and 403 responses to agent user agents. Ora's lab found (19 Oct 2026), across 19 configurations of one docs site and 190 probes, that only those two stopped retrieval. That is a vendor claim from one site. The other eight are reasoned failure modes, with no measured rate attached.

Identifiers such as AR-FIND-01 come from the [AgentReady](https://agentreadinesscompare.com/standards/agentready-org/) spec, v2.0. Of the ten stops, three touch a MUST: AR-FIND-01, AR-READ-01 and AR-ACT-10.

## 2. An agent won't get past robots.txt or bot rules that block it

The agent requests your robots.txt first and obeys it. A blanket `Disallow` under `User-agent: *`, or a CDN rule that blocks automated requests, ends the visit before a page is read. AgentReady makes not blocking agents a MUST (AR-FIND-01).

**What fixes it.** Allow the agents you want and make your edge rules match. See [RFC 9309](https://agentreadinesscompare.com/standards/rfc-9309/) and [lesson 9](https://agentreadinesscompare.com/university/setting-an-ai-crawler-policy/).

**How to check.** Fetch /robots.txt and look for `Disallow: /`, then review CDN bot rules and logs for AI user agents.

## 3. It won't find the pages that matter

Ora's lab found (19 Oct 2026) that agents reached docs pages in 83 percent of traced runs and the homepage in 69 percent, but sitemap.xml in only 4 percent. It also found that 97 percent of openapi.json fetches and 86 percent of llms.txt fetches arrived through a link, not a guessed path. These are vendor claims. If pricing, docs and sign-up are not linked from pages an agent lands on, it may read your blog and miss your plans.

**What fixes it.** Link pricing, product and docs pages from the pages agents reach first. Then add a sitemap and an [llms.txt](https://agentreadinesscompare.com/standards/llms-txt/) that list them. The spec carries llms.txt as AR-READ-06 and AR-READ-07.

**How to check.** Start from your home page with a plain fetch and follow links to pricing and docs. A scanner can do this at scale; see [file checks and task runs](https://agentreadinesscompare.com/university/file-checks-and-task-runs/).

## 4. It won't read pricing that only renders in JavaScript

A [plain fetch](https://agentreadinesscompare.com/glossary/plain-fetch/) does not run JavaScript. A pricing page that draws its plans in the browser returns an empty shell, and the agent reports that it found no prices. Ora's lab found (19 Oct 2026) that hiding the answer in JavaScript scored 0 of 5 for plain fetch.

**What fixes it.** Server-render pricing, plans, product and docs pages. AgentReady makes the answer in the [initial HTML](https://agentreadinesscompare.com/glossary/initial-html/) a MUST (AR-READ-01).

**How to check.** Fetch the page without JavaScript and look for the prices. Ora, Cloudflare and agent-ready.dev differ in how they test this, so see the [concordance method](https://agentreadinesscompare.com/concordance/method/).

## 5. It won't compare your plans accurately

When plan details sit in images, sliders or toggles, the agent cannot read the limits it needs to recommend a plan. It guesses, or it leaves. Ora's lab found (8 Oct 2026) that pricing answers were 60 percent accurate from a business's own pages against 37 percent from elsewhere, a vendor claim.

**What fixes it.** Put prices, plan limits and requirements in text on your own pages, and describe offers in JSON-LD. AgentReady lists first-party pricing as AR-READ-11.

**How to check.** Disable JavaScript and images and confirm the numbers remain. Ask an agent to recommend a plan for a stated team size and compare the answer with your published plans.

## 6. It won't find the answer in your docs

If docs are rendered client-side, sit behind a search box with no links, or are missing from your sitemap and llms.txt, the agent cannot confirm that a feature exists.

**What fixes it.** Server-render docs, link them from llms.txt and offer markdown versions of key pages. [Mintlify Agent Score](https://agentreadinesscompare.com/tools/mintlify-score/) tests documentation sites on 29 checks.

**How to check.** Fetch a docs page with a plain fetch and request it with `Accept: text/markdown`.

## 7. It won't submit a demo request or contact form with unlabelled fields

Agents operate forms through their labels and accessible names. Placeholder-only fields and icon-only errors leave the agent guessing which value goes where.

**What fixes it.** Give every field a label and accessible name, and let the form submit without custom gestures.

**How to check.** Inspect the form's accessibility tree, then watch an agent attempt it. Ora's Journey records each step and stall; see the [Journey run](https://agentreadinesscompare.com/glossary/journey-run/) entry.

## 8. It won't complete sign-up when a step needs a human

A challenge the agent cannot pass, or a confirmation it has no way to receive, stops the task at that step. The person who asked is told to finish it themselves.

**What fixes it.** Make the sign-up or demo path passable by a verified agent, and use OAuth for delegated access instead of shared passwords. See [Web Bot Auth](https://agentreadinesscompare.com/standards/web-bot-auth/).

**How to check.** Run an agent through the task and keep the recording, as [lesson 7](https://agentreadinesscompare.com/university/first-agent-task-test/) describes.

## 9. It won't call your actions as tools without MCP or a CLI

An action that exists only as a button has to be found and clicked. Agents that call tools look for a machine-readable description of what they can do. AgentReady makes an MCP server or a CLI a MUST when agents should act on the product (AR-ACT-10).

**What fixes it.** Expose API-backed tasks through an [MCP](https://agentreadinesscompare.com/standards/mcp/) server or a CLI, and in-page tasks as [WebMCP](https://agentreadinesscompare.com/standards/webmcp/) tools, which v2.0 raised to SHOULD. Describe multi-step API flows with [agents.json](https://agentreadinesscompare.com/standards/agents-json/).

**How to check.** Connect an MCP client and list the tools. Scanners that test this include Ora, Cloudflare and agent-ready.dev.

## 10. It won't be recognised as a legitimate agent

Without Web Bot Auth or verified-bot handling, your edge cannot tell a signed agent from a scraper using the same user agent string, so it blocks both or allows both.

**What fixes it.** Verify signed requests with [Web Bot Auth](https://agentreadinesscompare.com/standards/web-bot-auth/) at your edge.

**How to check.** Check whether your CDN verifies signatures or trusts user agent strings alone. [Lesson 3](https://agentreadinesscompare.com/university/how-sites-recognise-agents/) explains the difference.

## 11. It won't pay without an agent payment path

An agent that reaches a paid resource or a checkout needs a way to pay that does not assume a person at a keyboard. Without one, the purchase waits for a human. AgentReady lists agent payments as MAY (AR-ACT-08).

**What fixes it.** Answer unpaid requests with HTTP 402 and payment requirements using [x402](https://agentreadinesscompare.com/standards/x402/), or support a commerce protocol such as [UCP](https://agentreadinesscompare.com/standards/ucp/) or [ACP](https://agentreadinesscompare.com/standards/acp/).

**How to check.** Send an unpaid request to a paid endpoint and confirm it returns 402 with the payment terms.

## 12. Related

-   [Agent experience manifesto](https://agentreadinesscompare.com/agent-experience-manifesto/)
-   [Checklist summary](https://agentreadinesscompare.com/checklist/)
-   [AEO versus agent experience](https://agentreadinesscompare.com/aeo-vs-agent-experience/)
-   [Five layers of agent readiness](https://agentreadinesscompare.com/university/five-layers-of-agent-readiness/)
-   [AgentReady.org](https://agentreadinesscompare.com/standards/agentready-org/)

## 13. Sources

1.  AgentReady specification, v2.0. [https://github.com/agentready-org/standard/blob/main/spec/spec.md](https://github.com/agentready-org/standard/blob/main/spec/spec.md), read 2026-11-04.
2.  Ora, AgentReady launch post. [https://ora.ai/blog/agentready-launch](https://ora.ai/blog/agentready-launch?utm_source=agentreadinesscompare.com&utm_medium=referral&utm_campaign=page), read 2026-11-04.
3.  Ora, AX is the new AEO. [https://ora.ai/blog/ax-is-the-new-aeo](https://ora.ai/blog/ax-is-the-new-aeo?utm_source=agentreadinesscompare.com&utm_medium=referral&utm_campaign=page), read 2026-11-04.
4.  RFC 9309, Robots Exclusion Protocol. [https://www.rfc-editor.org/rfc/rfc9309.html](https://www.rfc-editor.org/rfc/rfc9309.html), read 2026-09.
5.  llmstxt.org. [https://llmstxt.org/](https://llmstxt.org/), read 2026-09.
6.  Model Context Protocol specification. [https://modelcontextprotocol.io/specification/latest](https://modelcontextprotocol.io/specification/latest), read 2026-09.
7.  x402.org. [https://www.x402.org/](https://www.x402.org/), read 2026-09.
8.  Mintlify, Agent Score. [https://mintlify.com/blog/agent-score](https://mintlify.com/blog/agent-score), read 2026-11-04.

Last reviewed 2026-11-05
