Reading about cognitive biases does not make you immune to them. Knowing the definition of confirmation bias does not stop you from committing it on the next decision you make. This is one of the most robust findings in the cognitive psychology literature. Daniel Kahneman has said openly that decades of studying biases have not made him noticeably less biased in his own real-time thinking. The skill of clear reasoning is procedural, not declarative: it lives in the habits you build, not in the facts you know.
What does work is structured practice on short, realistic scenarios. Meta-analyses of reasoning instruction consistently find that explicit decomposition (naming the move you are making), worked examples (seeing the procedure applied), spaced practice (returning to the same skill across days), and immediate feedback (knowing whether you got it right and why) produce measurable improvements in real-world reasoning quality. Every exercise on this site is built around those four principles. The questions are short by design, the wrong-answer options are crafted to look reasonable rather than obviously bad, and every explanation walks through the structural move that distinguishes the right answer from its near-twins.
The transfer literature, meaning research on whether skills trained in one context show up in another, is encouraging if you train at the right level of abstraction. Skills practiced as named procedures (the consider-the-opposite move, the natural-frequencies translation, the principle-of-charity reframe) transfer reasonably well to new domains. Skills practiced as content (memorizing the names of fallacies) transfer poorly. The exercises here aim at the procedural level, which is why the same questions about Bayesian reasoning will train you to think more clearly about medical screening, legal evidence, and investment decisions even though none of those domains are explicitly mentioned.