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

# Playground

> Send test requests through your gateway, tune parameters, and inspect the raw request and response.

## What this is for

**Playground** is an in-dashboard chat UI that sends real requests through your activated gateway. Use it to smoke-test a provider after connecting it, debug guardrails, validate routing rules, or compare models side-by-side without writing any client code.

Every playground request flows through the gateway exactly like a production call, so it's also the fastest way to verify a change to providers, models, routing, or guardrails.

<Warning>
  **Playground requests originate from the Guardway platform, not your browser.** A **Healthy** pill in the dashboard only proves the gateway can reach the Guardway control plane outbound — it does **not** prove the platform can reach the gateway inbound. For Playground to work, the gateway must be **published and reachable over the internet** at the URL you connected (a public hostname / port, a tunnel, or a reverse proxy with TLS). A gateway running only on a private LAN will register and heartbeat fine, but every Playground request will fail to connect.
</Warning>

## Options

| Control           | What it does                                                   |
| ----------------- | -------------------------------------------------------------- |
| **Provider**      | Filters the model list to one provider.                        |
| **Model**         | Any enabled chat model from the selected provider and gateway. |
| **System prompt** | Optional textarea. Sets the system message.                    |
| **Temperature**   | `0`–`2`. Default `1`. Lower is more deterministic.             |
| **Max tokens**    | Upper bound on completion length.                              |
| **Top-P**         | Nucleus sampling.                                              |
| **Streaming**     | Toggle. When on, tokens arrive incrementally.                  |

The right rail surfaces **Request JSON** and **Response JSON** tabs so you can copy the exact payload into your own client.

## Requirements

* At least one **Healthy** gateway — see [Activate](/guardway-gateway/activate).
* At least one provider attached to that gateway with valid credentials — see [Providers](/platform/configuration/providers).
* At least one enabled model — see [Models](/platform/configuration/models).

If any of these is missing, the Playground shows an inline empty state with a link to the page that fixes it.

## How to configure

<Frame caption="Playground">
  <img src="https://mintcdn.com/fcguardwayai/rJTQ_bXDRs9Cgazf/images/screenshots/platform/playground/playground.png?fit=max&auto=format&n=rJTQ_bXDRs9Cgazf&q=85&s=262312c32e7564a5f0780ca7cf01cdd3" alt="Playground" width="1611" height="1147" data-path="images/screenshots/platform/playground/playground.png" />
</Frame>

<Steps>
  <Step title="Open Playground">
    Open **Playground** from the dashboard sidebar.
  </Step>

  <Step title="Pick a provider and model">
    Use the **Provider** dropdown to filter, then pick an enabled **Model**.
  </Step>

  <Step title="Set sampling parameters">
    Adjust **Temperature**, **Max tokens**, and **Top-P** as needed. Toggle **Streaming** on to see tokens arrive incrementally.
  </Step>

  <Step title="Send a prompt">
    Type into the chat box and press **Send**. The response renders in the chat pane and the **Request / Response JSON** tabs populate on the right.
  </Step>

  <Step title="Verify in Logs">
    Open [Logs](/platform/logs) — the request appears with its trace ID, latency, and cost. Click the trace to see every step the gateway took.
  </Step>
</Steps>

## Tips

<AccordionGroup>
  <Accordion title="Test guardrails">
    Enable a PII rule under [Configuration → Security](/platform/configuration/security#guardrails), then paste a prompt containing a fake SSN into the Playground. The gateway should block or redact it depending on the rule's action — and the trigger will show in the **Guardrails** tab of [Logs](/platform/logs).
  </Accordion>

  <Accordion title="Compare models">
    Open Playground in two tabs, pick different models, send the same prompt. The Traces view in [Logs](/platform/logs) lets you compare latency and cost after the fact.
  </Accordion>

  <Accordion title="Reproduce in your client">
    The **Request JSON** tab shows the exact payload sent to the gateway, so you can drop it into your own client unchanged.
  </Accordion>
</AccordionGroup>

## Related

* [Logs](/platform/logs) — every Playground request is logged and traced.
* [Configuration → Routing](/platform/configuration/routing) — rules that decide which provider serves a Playground request.