CoT step counter and planner
Chain-of-thought prompting works best when the number of steps fits the task. Too few and the model leaps to a conclusion it cannot support; too many and it wastes tokens and sometimes talks itself out of a right answer. This tool reads your task, estimates how many reasoning steps it genuinely needs, and writes a numbered scaffold so the model reasons in the right shape from the start.
How it works
The estimator looks for complexity signals in your task text: overall length, conditional words (if, unless, otherwise), comparison and ranking language, explicit multi-part requests, arithmetic, and stated constraints. It sums those into a score and maps the score to a complexity band — from trivial to highly complex — each with a recommended step range. It then generates a numbered scaffold where each step has a clear job (understand, decompose, analyze, synthesize, check). You can override the complexity if you know better than the heuristic, and the scaffold updates instantly.
Tips and notes
- Override when you know more. A short task can still be hard. If the estimate looks low, bump the complexity and the scaffold expands.
- Keep a checking step. The final step in every scaffold asks the model to verify its own answer — that one step catches a surprising share of errors.
- Do not pad simple tasks. If the tool says two steps, trust it; forcing a five-step scaffold onto a one-line lookup is counterproductive.
- Pair it with a reasoning trace. Combine this scaffold with a tagged reasoning block so the steps are both planned and inspectable.