# AgentReady requirements explained

How AgentReady's find, read and act essentials apply to different websites, what automation can check and where expert judgement is needed.

[Doğu Abaris](https://tityra.com/authors/dogu-abaris) · Published 2026-10-29

> Choose the checks that match what the website is meant to do. Record what you observed and which checks you left out.

## Choose checks for what the website does

AgentReady groups its essentials around what an agent needs to find, read and act. An editorial website needs to publish information that agents can find and understand. A service that lets agents transact also needs usable interfaces and appropriate permissions.

Choose checks based on what the website is meant to support. In your assessment, record the resources you checked and explain what you left out.

[How to audit a website for Agent Experience](/blog/agent-experience-audit)

## Find: discovery and access

Review available first-party guidance, links, sitemap entries and the stated crawl policy for the selected sample. Retrieval checks examine whether the tested client receives the expected public resource or encounters an error, redirect or challenge.

These checks can collect status codes, headers and discovery-file contents automatically. Review access restrictions against the intended audience. A challenge may be correct for a private customer area. If it appears on a public page, check the CDN or application rules before recommending a change.

- Declared discovery and crawl policy
- Actual responses for sampled public resources
- Redirect destinations and missing-page behaviour
- First-party guidance linked to the relevant resource

## Read: content and meaning

Examine whether the content has a clear semantic structure and enough context to be interpreted accurately. Checks can include headings, landmarks, links, tables, metadata and structured data, along with authorship, date, source and usage information. Compare the initial HTML with the rendered page to see whether both contain the information needed for the task.

Automation can identify missing fields, parse errors and differences between views. Review the actual page and task to judge whether an answer is complete or a usage statement fits its audience. A full accessibility assessment also needs manual testing.

## Act: only where the product supports it

When a product is meant to let agents take an action, the assessment may examine documented interfaces, schemas, authentication, permissions and what happens after an error. Keep those checks within the agreed access. For a site that only publishes information, focus on the checks for finding and reading it.

Check where the action takes effect. Can the agent distinguish preparing a request from submitting it? Does the owner still provide the required approval? Are errors and retry instructions usable? Confirm authorisation before testing production changes, purchases or operations involving private data.

## What the report should make clear

Read each observation with its affected resources, method, date and evidence. Label confirmed failures, inferences, unavailable measurements and checks outside the scope separately. Use only completed checks in score summaries.

Record the assessment date and the version of the framework or checks used. Agree a scope for any deeper investigation or retest.

## How Tityra uses the essentials

Tityra's AX audits check against the AgentReady essentials within an agreed website scope. The checks focus on content access and delivery, semantic structure, metadata, structured data, accessibility signals and source context. The broader checks for authentication, permissions and task completion apply to products designed to let agents act.

The report describes the checks performed on the agreed sample. It records the findings, evidence and recommendations. Use the Tityra workspace to discuss findings and track progress across projects. If you need independent verification of a fix, agree a follow-up assessment.

## Sources and further reading

- [AgentReady essentials](https://www.agentready.org/)

## About the author

### [Doğu Abaris](https://tityra.com/authors/dogu-abaris)

Doğu Abaris is a co-founder of Tityra and a software and data engineer. An open-source contributor to Wikimedia and MediaWiki projects since 2011, he writes about agent access, semantic structure and evidence-led website audits.

[Tityra audit methodology](https://tityra.com/audit-methodology) · [Blog](https://tityra.com/blog)