Secret command list

There is active discussion online about the so-called “secret command list” for working with ChatGPT. In reality, this is a convenient way to structure prompts and predefine the desired response format. It is not about hidden capabilities, but rather a system of prompts that allows control over style, depth, and logic without constant clarifications during the conversation.

The idea is simple and, in a way, quite traditional — similar to good engineering practice: first define the task, then set the parameters, and only then generate the result. This approach saves time and makes interaction with the neural network more predictable, especially when dealing with complex topics such as finance, investing, analytics, or content preparation.

A particularly interesting aspect is the division by explanation levels. For example, EL1 is suitable for very basic answers, when you just need to grasp the essence without details. EL10 provides a clear step-by-step structure, which is useful for learning or explaining complex topics in simple terms. EL20 adds depth and detail, while EL30 turns the answer into a full analysis with causes, consequences, and nuances that can serve as a basis for an analytical article or research.

Style control is equally important. Commands like /human, /concise, or /formal allow you to set the tone of communication right away. This is especially useful in business correspondence or when preparing publications where both meaning and form matter. The /fun command, on the other hand, adds lightness and humor — sometimes exactly what is needed to make dry explanations more engaging and understandable.

It is also possible to regulate the depth of thinking. The basic X1 provides a fast and direct answer without extensive reasoning. But when using X10 think or X20 think, the model performs deeper analysis, considering causes, possible consequences, and even alternative scenarios. This mode is especially valuable in analytics, where it is important not only to know “what is happening,” but also “why it is happening and what comes next.”

Another important aspect is the honesty layer. Commands like Killcritic or RealityCheck essentially enforce a stricter response mode, where the model avoids excessive smoothing and provides a more direct assessment. This can be useful when analyzing news, investment ideas, or controversial topics where it is important to see both strengths and weaknesses.

For clarity, here is everything in a list:
Explanation level:
🟠EL1 – very simple and поверхневое explanation;
🟠EL10 – clear step-by-step answer without complex terms;
🟠EL20 – detailed explanation with all nuances;
🟠EL30 – in-depth analysis with causes, consequences, and comparisons.

Response style:
🟠Command /human – natural, human-like text;
🟠Command /concise – short, to-the-point answer;
🟠Command /friendly – warm and friendly tone;
🟠Command /formal – strict business style;
🟠Command /fun – light response with humor.

Thinking depth:
🟠X1 – basic answer without deep reasoning;
🟠X10 think – deep analysis with explanation of causes;
🟠X20 think – maximum depth with analysis of causes, consequences, and different scenarios.

Honesty:
🟠Killcritic – direct and honest assessment without softening;
🟠RealityCheck – verification of ideas or news against reality.

Overall, this is not about hidden features but rather a logical way of working with a tool. As in any system, the more precise the input, the more accurate the output. ChatGPT behaves in a rather traditional way here — like a good specialist who performs better when given a clear task.

In practice, these commands turn interaction with a neural network into a controlled process. The user is no longer just asking questions but becomes more of a “conductor” who defines the tone, depth, and direction of the response. And there is clear value in that — especially for those who work with information daily and value not only speed but also the quality of the result.