Assumption Audit
Reasoning
Level 2 · Advanced reasoningList the key assumptions behind your reasoning before answering, noting how sensitive the answer is to each one.
Assumption Audit asks the model to explicitly list the assumptions its answer depends on before committing to a final answer, optionally noting which assumptions the conclusion is most sensitive to. Most wrong answers to non-trivial questions trace back to a hidden, unexamined assumption rather than a reasoning error in the steps that follow it — surfacing assumptions before answering makes them checkable instead of invisible.
When to use it
Use it for decisions where a hidden wrong assumption would meaningfully change the conclusion — forecasts, recommendations, anything built on data or context that might be incomplete or stale.
When to avoid it
Avoid it for tasks with no real uncertainty — a straightforward factual lookup or a fully-specified task has no meaningful assumption to surface.
Advantages
- Surfaces a wrong assumption before it silently determines the conclusion
- Makes it possible to challenge just the assumption, not the whole answer, when something turns out wrong
- Highlights which parts of a conclusion are load-bearing vs. incidental
Disadvantages
- Adds length before the actual answer
- A model can list plausible-sounding assumptions without genuinely checking their impact on the conclusion
Examples
Should we expand into the Brazilian market next year? Input: current revenue is 80% domestic, we have no local partnerships yet, a competitor entered 6 months ago. Before answering, list the key assumptions your answer depends on, and note which ones the conclusion is most sensitive to. Then give your final answer.
Pairs well with
Try it
Fill in the placeholders below and see the compiled prompt update live.
What you want to achieve.
The material to work from.
- 28
- Words
- 183
- Characters
- 0/2
- Placeholders filled
- 1 min
- Reading time
- Completeness0
- Clarity0
- Context0
- Specificity0
- Constraints0
- Format0
- Output0
- Constraints note0
- Success criteria0
- Scope focus100
- Verification step0
- Negative example0
- Confidence0
- Placeholder syntax100
- Markdown structure0
- Fill in the remaining placeholders.
- Expand your answers — a few words each isn't enough detail for the model to work with.
- Add background context so the model understands the situation.