To make requests to the Gork AI, you need to use the X (twitter) API.

An API (Application Programming Interface) is a set of functionalities made available to a client/user via a library (methods of a programming language) or a web service.

Setting up the API

To make a request, you need an API key. It will identify who is doing what (in this case, who is using Gork). To generate a key, you can go to this site, in the corresponding tab.

Setting up Voltask

Once you have obtained the API key, you need to prepare the HTTP Request block in Voltask to send a request:
the method is POST
the URL is https://api.x.ai/v1/chat/completions
the body looks like this: {"messages": [{"role": "user","content": "{{prompt}}"}],"search_parameters": {"mode": "auto","return_citations": true,"sources": [{ "type": "web", "country": "{{country}}" },{ "type": "x"}, {"type": "news", "country":"{{country}}"}]},"model": "grok-3-latest"}
With {{ prompt }} as the query sent to Gork and {{ country }} as the ISO 3166-1 alpha-2 country code (e.g., FR for France)
the header looks like this: {"Content-Type":"application/json","Authorization":"Bearer {{APIkey}}"}
With {{ APIkey }} being the API key obtained earlier.

For more details on how to fill the request body, here is the link to the documentation: https://docs.x.ai/docs/guides/live-search

Retrieving the Response

The API returns a JSON object. It is necessary to select only what is important for a user.

JSON (JavaScript Object Notation) is a textual data format used for data transmission. A JSON object is a string (text) that contains the data to be transmitted. These objects are enclosed in braces, and associate a key with a value (text, list, number, boolean value, or another JSON object), separated by colons (for example: {"fruits": ["apple", "banana"], "number": 2, "likes": true})

In Voltask

It is possible to start the search work from Voltask, using the From JSON block. Here is a schema for retrieving information from the JSON object:

{"$schema": "http://json-schema.org/draft-07/schema#","properties": {"choices": {"items": {"properties": {"finish_reason": {"type": "string"},"index": {"type": "number"},"message": {"properties": {"content": {"type": "string"},"refusal": {},"role": {"type": "string"}},"required": ["role","content","refusal"],"type": "object"}},"required": ["index","message","finish_reason"],"type": "object"},"type": "array"},"citations": {"items": {"type": "string"},"type": "array"},"created": {"type": "number"},"id": {"type": "string"},"model": {"type": "string"},"object": {"type": "string"},"system_fingerprint": {"type": "string"},"usage": {"properties": {"completion_tokens": {"type": "number"},"completion_tokens_details": {"properties": {"accepted_prediction_tokens": {"type": "number"},"audio_tokens": {"type": "number"},"reasoning_tokens": {"type": "number"},"rejected_prediction_tokens": {"type": "number"}},"required": ["reasoning_tokens","audio_tokens","accepted_prediction_tokens","rejected_prediction_tokens"],"type": "object"},"prompt_tokens": {"type": "number"},"prompt_tokens_details": {"properties": {"audio_tokens": {"type": "number"},"cached_tokens": {"type": "number"},"image_tokens": {"type": "number"},"text_tokens": {"type": "number"}},"required": ["text_tokens","audio_tokens","image_tokens","cached_tokens"],"type": "object"},"total_tokens": {"type": "number"}},"required": ["prompt_tokens","completion_tokens","total_tokens","prompt_tokens_details","completion_tokens_details"],"type": "object"}},"required": ["id","object","created","model","choices","usage","system_fingerprint","citations"],"title": "Generated schema for Root","type": "object"}

However, out of all this information, only the choices attribute is necessary. Unfortunately, this attribute is an array and cannot be interpreted directly from the From JSON block; JavaScript will need to be used, and Celestory allows to do it.

In Celestory

Once the choices array is retrieved, you can use the Run Javascript block to retrieve only the AI response. Here is a code that accomplishes this task:

const raw = celestoryPoints.get("answer", "");
let answer;
try {
    answer = typeof raw === "string" ? JSON.parse(raw) : raw;
    const content = answer?.[0]?.message?.content;
    if (content) {
        celestoryPoints.set("res", content);
    } else {
        console.error("Content not found in structure:", answer);
    }
} catch (e) {
    console.error("Error parsing or accessing content:", e, raw);
}
celestoryPoints.set("res", answer[0]["message"]["content"]);

This way, it is possible to retrieve the AI response and use it in other blocks.
Note that a response from the HTTP Request block may take a few tens of seconds.

Updated on: 23/05/2025