Build a complete workflow with a thinking model. Return JSON {"task":"problem with restrictions","prompt":"a request that provokes deep reasoning","reasoning_request":"how you ask to show the steps and choice between options","verification_plan":"how you check the logic and assumptions of the answer","mistake_avoided":"what typical mistake from the module are you deliberately NOT making"}
The outcome of the module is a four-step work cycle. First: set a task with context and restrictions (three weeks, one day off, don’t disrupt work, limited budget). Second: call for a step-by-step reasoning and an explicit comparison of options - “themselves versus the movers.” Third: check the answer - pull out the assumptions, recalculate the key step, ask for a counterexample and self-criticism. Fourth: decide. Keep the rake of the module in mind: do not turn on the thinking model for trifles; do not confuse the length of the answer with logic; do not trust a confident tone without checking; always give restrictions, otherwise you will get general advice; follow the leading questions - the model plays along. The master’s main trick: keep two requests - one “solve the problem”, the second “now find the weak point in your own solution.” The discrepancy between them is the most valuable thing: there is something hidden there that you would otherwise miss. It is this cycle that distinguishes someone who uses the thinking model consciously from someone who simply presses a button and hopes.
Unlock access to submit solutions for instant AI review.