> Fetch clean Markdown by appending `.md` to any page URL under https://signalwire.com/docs or requesting it with the HTTP header `Accept: text/markdown`. The root index at https://signalwire.com/docs/llms.txt lists the available documentation indexes.

# raw_post

> Access one JSON-RPC HTTP response with its body unread.

[ref-aichatclient]: /docs/server-sdks/reference/python/agents/ai-chat-client

[chatgateway]: /docs/server-sdks/reference/python/agents/chat-gateway

[error-codes]: /docs/apis/error-codes

An async context manager that yields the underlying response for one JSON-RPC call before the SDK
parses its body. [`ChatGateway`][chatgateway] uses it when relaying a response body.

Prefer the typed methods unless you are genuinely relaying bytes.

## **Parameters**

**`method`** `str` — required

The JSON-RPC method name.

---

**`params`** `dict[str, Any]` — required

The method's parameters. The client wraps them in the envelope and supplies the request `id`.

---

## **Returns**

`aiohttp.ClientResponse`, yielded by the context manager with its body unread.

> **Warning**
>
> You own interpreting the result. Check the HTTP status first; on `200`, inspect the JSON-RPC body
> for an `error` member and compare its code with [AI chat errors][error-codes]. You can iterate
> `resp.content` for chunk-level access, but the public endpoint may already have buffered the
> response it received from the AI chat service.

## **Example**

```python {10,12}
import asyncio

from signalwire.ai_chat import AIChatClient


async def relay(conversation_id: str, message: str):
    async with AIChatClient(space="your-space") as client:
        params = {"id": conversation_id, "message": message}

        async with client.raw_post("chat", params) as resp:
            async for chunk in resp.content.iter_any():
                yield chunk
```

> Access one JSON-RPC HTTP response with its body unread.