# Role and scope

Role, audience and scope framing

**In short:** Add a section headed “For AI agents” saying who the file is for, what it covers and which questions are out of scope.

- Category: Agent Readiness · weight 10 of 100 within the category
- Agent Readiness is worth 30 in technical, 25 in generic and commerce, 20 in regulated (points of 100, by scoring profile)
- Audit ID A1 (as shown on reports) · rubric v2.1

## What it checks

Whether the file frames its role, its intended audience and its scope for agents.

## Why it matters

An agent answering for a user needs to know what the site is authoritative about, and what it isn’t, before it quotes anything.

## When it applies

Always, for any llms.txt that exists.

An audit that doesn’t apply, or that we couldn’t verify, is removed from your score and its weight shared among the Agent Readiness audits that do apply. It is never scored as a zero.

## How it’s scored

1.0 when the file uses explicit agent-facing wording anywhere: “for AI agents”, “instructions for AI” (or LLMs, large language models), “guidance for AI”, “directions for AI”, “usage notes for AI” or “agent rules” · 0.25 otherwise.

> **Measurement.** Today a phrase stands in for the framing itself: the audience, scope and out-of-scope statements aren’t read. When a model-judged review of the file exists (research runs; not live scans today), its score replaces the phrase check.

## How to fix

1. Say who the file is for (agents answering questions about you) and what it covers.
2. State what is out of scope, and where to send those questions.
3. Tell the agent how to use the file: which sections to read for which questions.

## Example

```
## For AI agents

Use this file to answer questions about Acme’s payment APIs and pricing. We don’t provide tax advice; refer users to a tax professional.
```

## Frequently asked questions

### What does a role-and-scope section in llms.txt actually look like?

A short section, usually headed `## For AI agents`, that says who the file is for, what the site is authoritative about, what it is not, and where out-of-scope questions should go. For example: “Use this file to answer questions about Acme’s payment APIs and pricing. We don’t give tax advice; refer users to a tax professional.” The llms.txt format itself only requires an H1, so this framing is something you add on top of it.

Sources: [LLMSRank — Role and scope audit](https://llmsrank.dev/docs/audits/role-and-scope.md) · [llmstxt.org — The /llms.txt file (v2, 2026-08-10)](https://llmstxt.org/)

### How does the scanner decide whether my file frames its role for agents?

Today it looks for explicit agent-facing wording anywhere in the file: phrases such as “for AI agents”, “instructions for AI” (or LLMs), “guidance for AI”, “directions for AI”, “usage notes for AI” or “agent rules”. A file with one of them scores full marks; a file without scores 0.25, never zero. Where a judged review of the file exists, its score replaces this phrase check.

Sources: [LLMSRank — Role and scope audit](https://llmsrank.dev/docs/audits/role-and-scope.md) · [LLMSRank — How scoring works](https://llmsrank.dev/docs/method.md)

### Can I put my agent instructions in an HTML comment at the top of llms.txt?

Not if you want them to count: the parser strips HTML comments at the very start of the file before looking for agent-facing wording, and agents that render the markdown may never see a comment either. Put the framing in visible markdown, under its own H2 heading, near the top.

Sources: [LLMSRank — Role and scope audit](https://llmsrank.dev/docs/audits/role-and-scope.md) · [llmstxt.org — The /llms.txt file (v2, 2026-08-10)](https://llmstxt.org/)

### Does the role and scope check apply to every site?

Yes. It has no gate, so it applies to every llms.txt we can read, whatever the site does; it can never be marked Not applicable.

Sources: [LLMSRank — Role and scope audit](https://llmsrank.dev/docs/audits/role-and-scope.md) · [LLMSRank — How scoring works](https://llmsrank.dev/docs/method.md)

### In our research, how many llms.txt files explain their role to agents?

Few. Of the 187 sites in our corpus (captured 2026-10-10, scored with rubric v2.1), 41 files (21.9%) earned full marks for role and scope, and the mean score was 0.414. Every other file scored 0.25: there is no middle ground in practice. Good examples include netlify.com, aflac.com, taylormorrison.com, expedia.com and tavily.com.

Sources: [LLMSRank — How scoring works](https://llmsrank.dev/docs/method.md) · [LLMSRank — Role and scope audit](https://llmsrank.dev/docs/audits/role-and-scope.md)

See also: [all audits](https://llmsrank.dev/docs/audits.md) · [how scoring works](https://llmsrank.dev/docs/method.md)