> ## Documentation Index
> Fetch the complete documentation index at: https://docs.koalr.ai/llms.txt
> Use this file to discover all available pages before exploring further.
# Agent Experience
> Agent Experience Optimisation (AXO): whether AI crawlers can reach and understand your site, scored across crawler access, schema markup and llms.txt.
The technical side of AI readiness. If crawlers are blocked, AI literally cannot cite you, and this is where you catch that.
## The AXO score
One number, 0 to 100, weighted across three parts:
| Part | Weight | What it measures |
| --------------------- | ------ | --------------------------------------------------------------------- |
| **Schema Markup** | 40% | How many pages carry valid structured data, and how well it validates |
| **AI Crawler Access** | 35% | The share of AI bots your site allows in fully |
| **llms.txt** | 25% | Whether you serve an llms.txt file and how well-formed it is |
When one part has not been audited yet, the score is rescaled over the parts that have, so a missing audit reads as unknown rather than as zero.
## AI Crawler Access
Koalr tests robots.txt and server-level responses for every relevant bot (**Re-check** runs it again), grouped into:
* **AI Assistants**: bots that power conversational AI products.
* **AI Search**: bots that power AI-enhanced search engines.
* **AI Data / Training**: bots that crawl for training data or indexing.
Each bot shows its robots.txt rule, its server access and an overall **Allow** or **Disallow**. **Server Blocked** means the server itself refused the request, which robots.txt cannot fix. **Audit History** lists every check with how many bots were allowed and blocked.
Blocking a training bot is a legitimate choice. Blocking an assistant or search bot removes you from that platform's answers entirely. Decide deliberately.
## Schema Markup
Structured data (JSON-LD) that tells AI what a page is about. The tab lists the **schema types found** across your crawled pages and the types you are missing, with **Validation Checks** for each. **Generate Schema** produces the markup for a page (Organization, WebPage, Article, FAQPage and so on) with instructions for where to paste it, and **Verify Schema** confirms it is live once you have.
## llms.txt
A standardised file that tells language models what your site is about and what content is available. It was proposed by Anthropic and has been adopted by a growing number of AI platforms. **Generate llms.txt** builds one from your site: a one-line summary for AI models, then your key pages and what each is for. Publish it at the root of your domain and use **Verify llms.txt** to confirm it is being served.
## Actions
Every failing check becomes an action in the [Action Centre](/act/action-centre), and Kevin's Top Tips at the top of the page front the highest-priority ones.
Agent Experience describes the site itself, not how visible you are in AI answers. A perfect AXO score with a low Prompt Visibility Rate means AI can read you but is not choosing you yet; that is a content and authority problem, and the [GEO Audit](/analyse/geo-audit) and [Sources](/monitor/sources) are where to look next.