◇ Could this help me?
Yes. The underlying advice is sound and matches how strong reasoning models actually behave: specifying the objective, the context, the constraints, and the definition of done gives the model room to pick a better approach than a human-scripted procedure would, and asking a natively reasoning model to show its thinking really is redundant. Two qualifications the video overstates: a role or persona is not always waste, it still helps when the audience or voice of the output matters, and dropping step-by-step instruction works only when the definition of done is genuinely specific, otherwise a vague goal produces a confidently wrong answer with nothing to check it against. General practice.
Rewrite this prompt as a goal-focused prompt. Strip out any role play, step-by-step procedure, and instruction to show reasoning, and restructure it into four parts: what I am trying to achieve, the relevant context including what is working and what is not, what is explicitly out of bounds, and what good and done look like. Do not tell the model how to work. Then point out any place where my definition of done is too vague to check the answer against, and ask me the question that would sharpen it.