AI Agent Website Test: Is Your Site Ready for Agents to Work With?
AI agents don't use the web the way people do — they work through protocols. Index Harmony measures your site's reading and understanding layers plus its agent-native protocol readiness, from AGENTS.md to MCP signals, and reports every gap with a ready-made fix.
No sign-up · 30–90 seconds · 100+ checks
How do AI agents use websites?
An agent working with your site is a three-step chain: it reads the page, understands what you offer, then acts on the user's behalf. The chain breaks at its weakest link:
1 · Read
Agents usually fetch your page without executing JavaScript. If your content isn't in the HTML, or bots are blocked, the chain breaks at step one.
2 · Understand
Structured data, heading hierarchy, AGENTS.md and llms.txt decide how an agent answers "what does this site offer, and on which page?"
3 · Act
To do business with you, agents need machine-facing doors: .well-known protocol signals like MCP, api-catalog and agent-skills are the addresses of those doors.
Index Harmony currently measures the first two layers plus the protocol readiness of the third; it does not simulate agent behaviour such as filling forms or completing purchases.
What do we check?
AGENTS.md
Do you have an AGENTS.md that tells agents how to use your site, reachable from your root directory?
Markdown negotiation
Do your pages serve a markdown version on request (Accept header, .md variants)? The most efficient reading format for agents.
MCP and .well-known signals
Do you publish agent-native discovery endpoints like .well-known/api-catalog and agent-skills?
Link header discovery
Do your HTTP Link headers announce machine resources such as llms.txt? Agents should find them without parsing HTML.
Structured data
Are your JSON-LD schemas (Organization, Product, FAQ) complete enough to feed the agent's "understanding" layer?
Who is it for?
E-commerce sites
Let shopping agents read and compare your products; start becoming an agent-ready site before your competitors do.
SaaS and API providers
Make your service discoverable to agents: api-catalog and AGENTS.md are your machine-facing storefront.
Booking and service sites
Keep appointments, menus and prices in a structure agents can read correctly.
Agencies and developers
Measure your clients' AI agent compatibility with a single report and apply fixes from ready-made instructions.
Frequently asked questions
What is an AI agent website test?
A free test that measures whether your website is ready for AI agents to work with: bot access, JavaScript-free readability, AGENTS.md, markdown support, structured data and MCP/.well-known protocol signals, scored in a single report.
What is the difference between an AI bot and an AI agent?
A bot is a crawler that reads and indexes content; an agent is software that tries to complete tasks on a user's behalf — researching, comparing, initiating actions. Agents need everything bots need, plus machine-facing discovery and protocol signals.
Does the test try filling forms or making purchases on my site?
No. Index Harmony does not simulate agent behaviour; it measures the reading and understanding layers, and for the action layer it checks protocol-readiness signals such as MCP, api-catalog and agent-skills.
Why do AGENTS.md and llms.txt matter?
Both are standards that describe your site to machines: llms.txt offers a content map, while AGENTS.md tells agents how to use the site. The report checks the presence and format of both.
What should I do if my score is low?
Every finding ships with a ready-made fix prompt — use "Copy to AI tool" to paste it into Claude Code, Cursor and similar tools. On WordPress, the Index Harmony plugin fixes most issues in one click.
Measure your site's agent readiness now
Enter your URL — we'll report your reading, understanding and protocol layers within 90 seconds.