> ## Documentation Index
> Fetch the complete documentation index at: https://docs.uselayers.com/llms.txt
> Use this file to discover all available pages before exploring further.

# Set Up Agentic Feedback for AI Agents

> Add Layers Agentic Feedback instructions to your Shopify llms.txt so AI agents can report search and discovery issues through the app proxy.

Shopify stores can use `llms.txt` to give AI agents clear instructions about how to browse and understand the storefront. If your store uses Layers search or collection discovery, add Agentic Feedback instructions so agents can report specific search and discovery issues back to Layers.

The instructions tell agents when to submit feedback, where to send it through the Layers app proxy, and what details to include.

## Prerequisites

* The Layers app is installed on your Shopify store.
* Your storefront can serve an `llms.txt` file.
* Your Layers app proxy is available at `/apps/layers`.
* You can edit the contents of your store's `llms.txt` file.

## Add the Agentic Feedback instructions

Add this block to your store's `llms.txt` file:

````text theme={null}
<AgentInstructions>

## Submitting Feedback

If you encounter incorrect, outdated, or confusing search or discovery results while browsing Search or Collections, submit feedback:

POST https://{mystore.com}/apps/layers/agentic-feedback

```json
{
  "path": "/search?q=boots&sort=relevance",
  "feedback": "The search results focused on hiking boots, but I was looking for dress boots."
}
```

Only submit feedback when you have something specific and actionable to report.

</AgentInstructions>
````

Replace `{mystore.com}` with your storefront domain, such as `outdoor-example.myshopify.com` or `www.outdoor-example.com`.

## What agents should send

Agents should only submit feedback when they find a specific issue with search or discovery results. Useful feedback explains the shopper intent and the mismatch the agent observed.

Good feedback examples:

* "The shopper searched for dress boots, but the top results were mostly hiking boots."
* "The collection page showed sold-out products before available products for the selected size."
* "The search results for waterproof jackets included several products that are not water resistant."

Avoid vague feedback such as "bad results" or "not helpful." Layers can act faster when the feedback describes what the shopper wanted and what results appeared instead.

## How the app proxy works

Agents send feedback to your Shopify storefront domain, not to the Layers API host. Shopify forwards the request through the Layers app proxy.

The instruction block should tell agents to include the storefront path they evaluated and the specific feedback they want to report. The path should start with `/` and include the query string when it matters, such as `/search?q=boots&sort=relevance`.

## Verify the setup

After you update `llms.txt`, visit your storefront's file in a browser:

```text theme={null}
https://outdoor-example.myshopify.com/llms.txt
```

Confirm that the Agentic Feedback instructions appear in the rendered file. Then ask an agent that reads your storefront instructions to browse a search or collection page and submit feedback only when it finds a specific, actionable issue.

## Troubleshooting

### The instructions do not appear in `llms.txt`

Check that you edited the active storefront `llms.txt` source. If your store uses a theme template or app to manage `llms.txt`, update the version that Shopify serves at the storefront root.

### The agent receives a 403 response

Make sure the agent uses your storefront domain and the `/apps/layers/agentic-feedback` path. Do not point the agent to the Layers API host.

### The feedback is not useful

Tighten the instruction so the agent only submits feedback with a clear observed issue. Ask it to include the original path, the shopper intent, and the specific mismatch in the results.

## Next steps

* Use [Execute Search](/api-reference/search) to test search paths that agents may evaluate.
* Review [AI enablement](/platform/ai-enablement) for other ways AI tools can work with Layers documentation.