---
title: "AI Content Signals"
description: "Publish AI usage preferences through robots.txt directives."
canonical_url: "https://nuxtseo.com/docs/ai-ready/guides/content-signals"
---

> Index of every page on this site: [llms.txt](https://nuxtseo.com/llms.txt)

Content Signals in `robots.txt`{lang="txt"} declare how you want AI systems to use your content.
They rely on the receiving system to honor them. They do not restrict access to a page.

## Standards

The module can emit two directive formats:

**[Content-Signal](https://contentsignals.org/)** (values: `yes`/`no`)

- `search`: indexing/snippets
- `ai-input`: RAG, grounding, AI search
- `ai-train`: model training/fine-tuning

**[Content-Usage](https://ietf-wg-aipref.github.io/drafts/draft-ietf-aipref-attach.html#name-robots-exclusion-protocol-c)** (values: `y`/`n`), opt-in

- `train-ai`: foundation model training

See [Nuxt Robots AI Directives](https://nuxtseo.com/docs/robots/guides/ai-directives) for full documentation.

## Default

With no `contentSignal`{lang="ts"} option, the module adds this line to the `User-agent: *`{lang="txt"} group:

```txt [robots.txt]
Content-Signal: search=yes, ai-input=yes
```

The default allows search and AI input. It states no preference on AI training, so the line leaves out `ai-train`{lang="txt"}.
Under the [Content Signals](https://contentsignals.org/) spec, a missing signal neither grants nor restricts that use.

The module adds no default line when a robots group or a merged `robots.txt`{lang="txt"} file already sets `Content-Signal`{lang="txt"}.
That policy stays the only one.

Set `contentSignal: false`{lang="ts"} to send no `Content-Signal`{lang="txt"} line.

## Response headers

Markdown conversion preserves a `Content-Signal`{lang="txt"} header from the source HTML response.
An explicit header on the Markdown route takes precedence.
Disabling automatic robots directives does not remove headers supplied by your application.

[Cloudflare Markdown for Agents](https://developers.cloudflare.com/fundamentals/reference/markdown-for-agents/#content-signals-policy) also preserves an origin policy header.
If the origin sends no header, the converter adds `ai-train=yes, search=yes, ai-input=yes`{lang="txt"}.
Configure an origin header when you need the same preference across both conversion paths.

## Prerendered policies

If your HTML response sends a `Content-Signal` header, prerendering preserves it on HTML and Markdown twins.
Explicit Markdown route policies and static header removals take precedence.
This preserves your existing headers. It does not create HTTP policy headers from `contentSignal` configuration.

[Cloudflare](https://cloudflare.com) builds write these policies into `_headers`.
Shared route policies stay shared. Specific policies replace broader policies.
Other static hosts need their own response header configuration.

[Cloudflare allows 100 static header rules](https://developers.cloudflare.com/workers/static-assets/headers/).
If optional discovery rules exceed this limit, the build drops them and retains policy rules.
If policy rules exceed the limit, the build fails. Use fewer per-page policies or a shared policy.

## Configure

Set each use you want to allow:

```ts [nuxt.config.ts]
export default defineNuxtConfig({
  aiReady: {
    contentSignal: {
      aiTrain: true,
      search: true,
      aiInput: true
    }
  }
})
```

The module adds these directive lines to the `User-agent: *`{lang="txt"} group. It merges them into a group you already define, so `robots.txt`{lang="txt"} keeps one wildcard group:

```txt [robots.txt]
Content-Signal: ai-train=yes, search=yes, ai-input=yes
```

Nuxt AI Ready requires [`@nuxtjs/robots`](https://nuxtseo.com/robots) 7.0.0 or later.

::warning
An object states the whole policy. The module sends `no` for each permission you leave out of the object.
For example, `contentSignal: { aiTrain: false }`{lang="ts"} sends `ai-train=no, search=no, ai-input=no`.
To keep the default and also refuse training, set all three permissions.
::

## Content-Usage

The module sends no `Content-Usage`{lang="txt"} line unless you set `contentSignal.contentUsage: true`{lang="ts"}.

```ts [nuxt.config.ts]
export default defineNuxtConfig({
  aiReady: {
    contentSignal: {
      aiTrain: true,
      contentUsage: true,
      search: true,
      aiInput: true
    }
  }
})
```

```txt [robots.txt]
Content-Usage: train-ai=y
Content-Signal: ai-train=yes, search=yes, ai-input=yes
```

Before you opt in, consider the trade-off:

- `Content-Usage`{lang="txt"} comes from the IETF AI Preferences [attachment draft](https://datatracker.ietf.org/doc/draft-ietf-aipref-attach/). The draft is not an RFC yet.
- The Lighthouse `robots-txt`{lang="txt"} audit reports `Content-Usage`{lang="txt"} as an unknown directive. This lowers the SEO score of every page.
- Lighthouse accepts `Content-Signal`{lang="txt"}. Its `ai-train`{lang="txt"} value states the same training preference.
- A comment form such as `# Content-Usage: train-ai=y`{lang="txt"} does not work. Parsers that follow the draft ignore comments.

## Selective Permissions

Allow search and AI input, and refuse AI training:

```ts [nuxt.config.ts]
export default defineNuxtConfig({
  aiReady: {
    contentSignal: {
      search: true,
      aiInput: true,
      aiTrain: false
    }
  }
})
```

```txt [robots.txt]
Content-Signal: ai-train=no, search=yes, ai-input=yes
```

## Sitemap

See the full [sitemap](https://nuxtseo.com/sitemap.md) for all pages.
