challenge-assumptions
/challenge-assumptions
Ujawnia ukryte założenia, sprawdza, co się stanie, gdy któreś okaże się fałszywe, i przeformułowuje problem przed rozwiązaniem.
- Autor
- aleknitka ↗
- Repozytorium
- agent-creative-thinking-skills ↗
- Licencja
- MIT
- Ostatnia zmiana repozytorium
- 13.11.2026
Ocena i bezpieczeństwo
- Pliki w paczce
- 8
- Skrypty
- 1
- Komentarze HTML
- 0
- Adresy URL w SKILL.md
- 0
Brak sygnałów. Repozytorium zmienione 13.11.2026, bez odwołań do wycofanych modeli i API.
MIT, otwarta: można używać i zmieniać, z zachowaniem informacji o autorze.
- 1 skrypt w paczce. Przejrzyj je przed instalacją, bo Claude może je uruchomić.
Skrypty: scripts/inventory.py
Co robi i kiedy
Ujawnia ukryte założenia, sprawdza, co się stanie, gdy któreś okaże się fałszywe, i przeformułowuje problem przed rozwiązaniem.
Wyzwalacze z opisu
Oryginalny opis (EN)
Make hidden beliefs explicit, test what happens if each is false, unnecessary, or reversed. Use when a problem feels overconstrained, every option looks impossible, or "we've always done it this way" / "policy says" dominates — inventory, test, and reframe before you solve.
Jak działa
- When to use
- When NOT to use (brief)
- Process
- Rules
- Going deeper
Podgląd SKILL.md
--- name: challenge-assumptions description: Make hidden beliefs explicit, test what happens if each is false, unnecessary, or reversed. Use when a problem feels overconstrained, every option looks impossible, or "we've always done it this way" / "policy says" dominates — inventory, test, and reframe before you solve. --- # Challenge Assumptions Most problems arrive pre-solved: a problem statement wrapped in beliefs that were never chosen, only inherited. "Onboarding *must* be a tutorial." "Approvals *have to* be manual." "Customers *won't* pay for that." If every option looks impossible, the constraint doing the blocking is usually a belief, not a fact. This skill makes those beliefs explicit, tests which ones actually carry the problem, and reframes the problem only where the evidence permits. It prevents two failures: accepting unnecessary constraints, and "challenging" constraints that are real and getting burned. ## When to use - The problem feels overconstrained; all obvious options are ruled out. - "We've always done it this way," "policy says," or "users need X" closes discussion. - A problem has been restated several times without getting smaller. - You suspect the stated problem is a solution in disguise. If unsure whether this is the right method, start with [thinking-router](../thinking-router/SKILL.md) — or see [references/when-not-to-use.md](references/when-not-to-use.md). ## When NOT to use (brief) - You need a new idea once the frame is sound → use `lateral-thinking` or `scamper`. - The bottleneck is missing facts about the world → research first; this skill will only label your ignorance. - There is a real engineering tradeoff to resolve → use `triz-contradictions`. - You are choosing among viable options → use `premortem` / `options-portfolio`. - The work is correctness-bound (proof, safety sequence, compliance audit) → use vertical / critical thinking. ## Process 1. **Intake — reuse before you ask.** In one sentence each: the stated problem, the goal behind it, the constraints you have been told are hard and who imposed them, what has already been tried, and appetite for changing scope (low / medium / high). Use existing context; ask only for gaps. Write the problem as an assertion, not a topic: "X must Y because Z." 2. **Inventory hidden beliefs.** List **8–15 beliefs**, phrased as checkable assertions, not themes. Cover inherited vs. proven, stakeholder vs. system, boundary, and continuity beliefs — including the ones buried in the problem statement itself. For each: type, source/owner, available evidence, confidence (low / medium / high), and what breaks if it is false. Use [references/inventory.md](references/inventory.md) and [scripts/inventory.py](scripts/inventory.py). Generate before judging. 3. **Pressure-test the drivers.** Rank beliefs by **impact × uncertainty** and test the top handful, not all of them. For each, ask three questions: *If false, what changes? If unnecessary, could the goal still be met? If reversed, what becomes possible?* Collect evidence, label every claim **fact / assumption / speculation**, and record what evidence would settle it. See [references/testing.md](references/testing.md). 4. **Reframe alternatives.** For the 2–4 load-bearing beliefs, apply the reframe patterns — **negate, relax, invert, re-scope, re-attribute** — and write **3 distinct problem statements** of the form "How might we [goal] without [assumption]?" Distinct means a different constraint is gone, not different wording. See [references/reframe-patterns.md](references/reframe-patterns.md). 5. **Small probes.** Choose the cheapest probe that could change your mind for each surviving uncertainty: existing data → desk check → one conversation → prototype → pilot. For each probe state the observable, the decision rule (what result means keep / kill / investigate), cost, and owner. Prefer evidence over debate; never test a safety- or compliance-critical assumption in production. Sizing and rules: [references/testing.md](references/testing.md). 6. **Decide next.** Recommend one of: (a) proceed under a reframed statement, (b) run a probe before deciding, (c) keep the constraint and record why, or (d) hand off — `lateral-thinking` for unexpected directions, `scamper` for variations, `first-principles` for rebuilding from constraints, `premortem` to harden a reframe. Confirm with the user before starting another method. Worked example: [references/examples.md](references/examples.md). ## Rules - State beliefs as assertions someone could disagree with. "Onboarding is a tutorial" — not "onboarding." - Separate inventory from evaluation: generate first, rank and test after. No premature "that's obviously true." - Test what blocks options. A belief that is uncertain but irrelevant is not worth a probe. - Never relitigate facts as if they were beliefs: physics, arithmetic, signed contracts, and observed behavior stay. Challenge *scope* and *policy*, then verify. - Reframes must remove or reverse a named belief. Rewording is not a reframe. - Label fact / assumption / speculation; do not present a plausible story as evidence. - One probe, one decision rule. A probe without a pre-committed interpretation is theater. - Honest provenance: name the lineage and call adaptations adaptations. See [references/provenance.md](references/provenance.md). ## Going deeper - Inventory — belief types, prompt list, capture table: [references/inventory.md](references/inventory.md) - Pressure-testing — falsify / unnecessary / reversal tests, evidence, probe sizing: [references/testing.md](references/testing.md) - Reframe patterns — negate, relax, invert, re-scope, re-attribute: [references/reframe-patterns.md](references/reframe-patterns.md) - Worked example — "onboarding must be a tutorial": [references/examples.md](references/examples.md) - When not to use + triage against lateral / SCAMPER / TRIZ / vertical: [references/when-not-to-use.md](references/when-not-to-use.md) - Provenance, lineage, and evidence limits: [references/provenance.md](references/provenance.md) - Inventory and test helper (offline): [scripts/inventory.py](scripts/inventory.py)
Pliki w paczce (8)
- SKILL.md 6 KB
- references/examples.md 8 KB
- references/inventory.md 8 KB
- references/provenance.md 7 KB
- references/reframe-patterns.md 8 KB
- references/testing.md 8 KB
- references/when-not-to-use.md 7 KB
- scripts/inventory.py 22 KB
Wersje
W bazie nie ma innych skilli o nazwie challenge-assumptions.
Migawka bazy z 13.11.2026 · wersja paczki f82cb5ccd9a8 (skrót SHA-256 całej paczki)
Pobrania w ostatnich 30 dniach: 0 · ranking
Zgłoś problem z tym skillem: szkodliwa treść, błędny opis albo prośba autora o usunięcie.
Podobne skille
- 90test-assumptions
Kwestionuje proponowane rozwiązanie przez wskazanie i sprawdzenie założeń o poważnych skutkach, gdy brief wydaje się zbyt pewny albo discovery ujawnia sprzeczności. - 81what-are-we-assuming
Sprawdza założenia, na których opiera się ocena lub plan, także gdy prośba jest raczej rozwiązaniem niż problemem. - 100decision-lens
Przeformułowuje trudny problem przed decyzją przez soczewki poznawcze z biblioteki 42, wskazuje ukryte założenia i przygotowuje brief decyzyjny z następnym krokiem. - 100plan-review
Audytuje plan przed implementacją: zakres, założenia, ryzyka i tryby awarii, zanim powstanie kod. - 95wicked-garden-deliberate
Sprawdza założenia przed pracą: przeformułowuje problem i pyta, czy zadanie jest postawione właściwie. - 82reframe
Przeformułowuje problem poza ujęciem, w jakim się pojawił, gdy ujęcie pochodzi z domyślności i każda opcja w nim przegrywa lub prośba nie nazywa celu.