Build the strongest possible case against your position, then use it to find the flaws in your thinking.

Prompt

I believe:
[STATE YOUR POSITION OR DECISION]

Step 1 - Steel Man: Construct the strongest, most charitable, most
intellectually honest argument AGAINST my position. Use the best
evidence, the most compelling logic, and assume the smartest
possible opponent.

Step 2 - Attack: Now use that steel man to attack my original
position. Point out flaws, blind spots, missing evidence, and
unintended consequences. Do not hold back.

Step 3 - Synthesis: Where does the truth likely lie? What should
I update in my thinking?

Example

Input:

I believe we should rewrite our monolith into microservices.

Output (abbreviated):

Steel Man Against:
The strongest case for keeping the monolith: Martin Fowler's
"monolith first" principle exists because microservices impose
a distributed systems tax -- network latency, data consistency,
deployment complexity, observability overhead -- that only pays
off at a scale most teams never reach. Your current team of 8
can deploy the monolith in minutes. With microservices, you'll
spend 6 months building infrastructure before shipping a single
feature. Amazon and Netflix had thousands of engineers before
they split.

Attack:
Your position assumes the monolith is the bottleneck. But your
deploy frequency is weekly -- is that because of code coupling
or because of process? If it's process, microservices won't fix
it. You'll have the same slow process times 12 services...

Synthesis:
Consider the "strangler fig" pattern instead: extract one bounded
context that genuinely benefits from independent deployment...