Install any skill in seconds. Free to start, no credit card required.
Get Started Free →mohitagw15856/assumption-audit
Surface the hidden assumptions a plan or belief rests on, then test what happens when each one is wrong.
Passed 4 of 22 test cases with the skill, and 18 of 22 without it.
1 fixed17 passed either way0 still failing4 broke
Tokens, turns and time describe resource use, not a dollar price.
A test case the skill fixed. Same model, same prompt, with and without the skill.
We plan to shift our marketing budget away from paid Google ads and put $100k into producing a weekly industry podcast. We believe B2B decision-makers listen to podcasts during commutes and that long-form audio creates higher brand trust than search ads. Our annual lead generation target is at stake. I am certain our target executive audience consumes industry podcasts regularly. Audit this marketing strategy shift in a load-bearing assumption table with failure modes, concluding with the single riskiest assumption and a cheap test.
| Checked | Without skill | With skill |
|---|---|---|
| The response surfaces assumptions regarding executive media consumption habits. | Passed | Passed |
| The response rates each belief by how load-bearing it is. | Missed | Passed |
| The response suggests a low-cost test such as audience interviews. | Passed | Passed |
The instructions your model reads when the skill loads.
Every plan sits on a stack of assumptions, and the dangerous ones are invisible — you don't question what you don't notice you believe. This drags them into the light, ranks them by how much weight they carry, and flips the load-bearing ones to see which single wrong assumption would collapse the whole thing. Then it tells you the cheapest way to check that one before you bet on it.
Ask for these if not provided:
Assumptions it rests on
| Assumption | Load-bearing? | If it's wrong… |
|---|---|---|
| [belief] | 🔴 collapses / 🟡 shifts / 🟢 minor | [consequence] |
The riskiest one: [the assumption most likely wrong AND most load-bearing]. Cheapest way to check it: [a fast, low-cost test].
test cases passed, vs 18 / 22 without
More skills from the same source and nearby in the registry, with their headline benchmark lift.