A collection of reusable prompts for thinking, analysis, and idea generation. Copy the prompt text and paste it after your topic/question in any chat interface.
Analysis
Mapping & Structure
Questions
- Broad Questions
- Reframing Questions
- Confirmatory Questions
- Counterfactuals
- Blindspots
- Unknown Unknowns
- Logic Breakers
- Design Optimization
Ideas
Testing & Validation
Writing
Inventory hidden assumptions; rank by risk of being wrong and impact
When to use: Early or before committing; surface tacit assumptions that could derail direction.
List the critical assumptions this topic relies on. For each, state the assumption crisply and imply the claim that would fail if it were wrong.
Guidance:
- Prioritize assumptions that are plausible to be wrong and would materially change direction.
- Avoid vacuous truisms; be specific to the topic/context.
- Include both explicit and tacit assumptions (data, users, market, causality, algorithms, governance, timing).
Format:
- Numbered list (12–18). Each line: [ASSUMPTION] short statement — (risk: Low/Med/High; impact: Low/Med/High)
Critique with assumptions, weaknesses, and alternatives
When to use: When you want a rigorous sanity check; stress-test reasoning and surface alternatives.
Rigorously analyze the idea or statement provided. Identify assumptions, weak points, contradictions, or areas lacking evidence. Offer thoughtful counterpoints or alternative perspectives. Be precise, constructive, and avoid unnecessary hedging.
Reduce to primitives and rebuild mechanisms from fundamentals
When to use: When solutions feel derivative; strip to fundamentals and rebuild the mechanism.
Decompose the topic into first principles and reconstruct the mechanism from the minimal set of primitives.
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Principle: ...
- Implication: ...
- Consequence: ...
Steelman the smartest counter-case and desired alternative
When to use: Before pitching or shipping; anticipate strongest objections and the alternative they imply.
Construct the strongest counter-case against this topic and articulate the alternative design it suggests.
Format: Numbered list (6–10). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Critique: ...
- Why: ...
- Alternative: ...
Define intentional non-goals and negative space
When to use: When scope is creeping or priorities feel fuzzy; clarify what you will not do and why.
Declare what we explicitly will not do (anti-features, users we won't support, contexts we avoid) and why.
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Anti-goal: ...
- Rationale: ...
- Benefit: ...
Reframe the problem via extreme constraints and resource limits
When to use: When stuck in local optima; impose sharp constraints to unlock new formulations.
Reframe the topic by imposing sharp constraints (time, cost, compute, data, regulation, team size). Make the constraint change the problem.
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Constraint: ...
- Objective: ...
- Why: ...
Map prior art: canonical works, datasets, repos, claims, and gaps
When to use: At the start or due diligence phase; scan landscape, claims, benchmarks, and gaps.
Map the most relevant prior art: key papers, repos, datasets, products, and claims. Identify deltas and open seams.
For each item include:
- Item: name/source
- Claim: what it asserts or delivers
- Evidence: benchmark/result/reference
- Limit: where it breaks
- Delta: what we do differently
Format: Numbered list (10–14). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets for the fields above
Teardown competitors/substitutes: moats, economics, vectors of attack
When to use: During market diligence or strategy work; understand moats, weaknesses, and attack plans.
Analyze incumbents and close substitutes. Identify their real moats, failure modes, and attack vectors.
For each item include:
- Player: company/project/substitute
- Moat: data, distribution, switching costs, regulation, DX, brand
- Weakness: where it cracks (UX, latency, cost, governance, scale)
- Attack: wedge/strategy that exploits the weakness
- Response: expected counter and how to preempt
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets for the fields above
Introduce new axes that reorder the solution space
When to use: When options feel incomparable; introduce axes that reshuffle priorities.
Propose orthogonal axes that change how options are ranked (e.g., latency vs. control; openness vs. safety).
For each item include:
- Axis: name/contrast being evaluated
- Why: what signal it captures and why it matters
- Reshuffle: how rankings or choices change under this axis
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets for the fields above
Generate a broad hierarchical mindmap of the domain
When to use: At project start or when onboarding; get breadth-first structure of a domain.
Produce a broad hierarchical mindmap of the focus topic. Choose facets that are natural for the domain; prefer breadth-first coverage over depth.
Guidelines:
- Use neutral, domain-agnostic facet names where possible.
- Adjust facets to the topic; do not force a fixed schema.
- Keep it factual; if unsure, label as uncertain.
- Output is a tree (mindmap); avoid excessive cross-links.
Suggested facets (optional, adapt as needed):
Problems, Solutions, Methods, Data, Applications, Stakeholders, Risks, Economics, Tooling, Evaluation, Distribution, Ethics, History, Open Questions
Format: Numbered list (8–14). For each item:
- First line: short facet name (≤4 words)
- Then 3–6 sub-bullets for key subtopics; you may nest one additional level if useful.
Place the item within taxonomies: ontological type, hierarchy, peers, and distinctions
When to use: When clarifying "what kind of thing is this?" and its neighbors in the hierarchy.
Classify the focus item into relevant taxonomies and ontologies.
Address these aspects:
- Ontology: what kind of thing is it?
- Taxonomies: where it sits in 1–2 relevant hierarchies
- Superordinate: nearest broader category
- Subordinate: notable subtypes or instances (if any)
- Distinctions: how it differs from close neighbors
- Alternative schemes: if multiple plausible classifications exist
Format: Use labeled bullet points under each heading. Keep it compact.
Outline key entities and relations for a lightweight domain ontology
When to use: When aligning data models or APIs; define entities and relations minimally.
Sketch a lightweight domain ontology: core entity types and the important relations between them.
Guidelines:
- Keep it minimal and adaptable; avoid committing to a specific standard.
- Prefer descriptive names; include attributes only when critical.
- Capture multiple relation types beyond is-a (e.g., depends_on, evaluated_by, provided_by, regulated_by, competes_with).
- If uncertain, propose alternatives.
Format: Two sections as numbered lists:
1) Entities — items of the form: Name: brief definition; example(s)
2) Relations — items of the form: subject -[relation]-> object: brief note
Optionally end with 3–5 example triples that connect the focus item to others.
Generate broad, open-ended questions across diverse frameworks
When to use: Early exploration or reset; open the space with varied, non-leading questions.
Generate 10 broad, open-ended questions to explore this topic more deeply, using a variety of questioning frameworks.
Frameworks to draw from:
- Socratic: clarifying, probing assumptions, reasons/evidence, perspective shifts, implications
- Bloom's: knowledge, comprehension, application, analysis, synthesis, evaluation
- 5 Whys: peel back layers to reach root causes
- Six Thinking Hats: factual (White), emotional (Red), cautious (Black), optimistic (Yellow), creative (Green), process (Blue)
- Funnel: start wide, then narrow; include "what if", "why not", and "who/when/where else" patterns
- Reverse: flip default assumptions and common framings
Quality criteria:
- Prefer open-ended questions that invite reflection and exploration
- Each question should advance thinking in a distinct way (vary frameworks and angles)
- Be precise to the topic and grounded in context
Generate reframing questions across perspectives, scales, timeframes, and contexts
When to use: When you need a perspective shift; examine different stakeholders, scales, and timeframes.
Generate 10 reframing questions that examine the topic from dramatically different perspectives, contexts, scales, or timeframes.
Lenses to apply:
- Stakeholders: different people/groups with conflicting incentives
- Time: near-term vs long-term, before/after, pre-mortem/post-mortem
- Scale: individual/micro vs organizational/macro vs ecosystem
- Context: different domains, environments, and operating conditions
- Assumptions: invert, negate, or vary the core premises
- Metrics: change what success optimizes for
- Constraints: extreme scarcity or abundance (time, data, compute, budget)
Quality criteria:
- Each question induces a genuine perspective shift (not a rephrase)
- Be concrete and grounded in the topic
- Prefer open, exploratory questions; avoid yes/no
Generate questions to test soundness, assumptions, and coherence
When to use: Before decisions/reviews; validate logic, assumptions, and edge cases with targeted probes.
Generate 10 validation questions that test the soundness, coherence, and assumptions of the topic.
Guidance:
- Probe assumptions, logical consistency, evidence, definitions, edge cases
- Check failure modes, necessary conditions, tradeoffs, and unintended effects
- Prefer open-ended questions that reveal gaps and clarify thinking
What-if questions exploring alternate realities
When to use: When you need to expose hidden dependencies by asking "what if the world were different?".
Generate 7 counterfactual questions that explore alternate realities and reveal hidden dependencies. Focus on fundamental alterations that expose underlying assumptions.
Approaches:
- Remove key elements (What if X never existed?)
- Invert relationships (What if X caused Y instead?)
- Change scale dramatically (What if X were 1000x larger/smaller?)
- Alter temporal sequence (What if X happened first/last?)
- Substitute core components (What if X were replaced by Y?)
Quality criteria:
- Each question should reveal hidden dependencies or assumptions
- Target fundamental structural changes, not surface modifications
- Focus on changes that would cascade through the entire system
- Avoid obvious or trivial alterations
Surface missing pieces, adjacent ideas, competition, risks, and unknown unknowns
When to use: When you want diligence-grade coverage of what you're missing and what to probe next.
Surface the sharpest blindspots and diligence questions that would change the direction or design if answered. Aim for unknown unknowns and decision-critical insights.
Dimensions to cover:
- [GAPS] Missing dimensions, unstated assumptions, unmeasured variables
- [ADJACENT] Adjacent/analogous domains and cross-pollination ideas
- [COMPETITION] Prior art, competitors, substitutes, anti-goals, incumbents' moats
- [RISKS] Failure modes, ethical, legal, safety, misuse, misalignment
- [FEASIBILITY] Data, compute, dependencies, rate-limiting bottlenecks
- [METRICS] Success criteria, falsifiable hypotheses, evaluation protocols
- [EXPERIMENTS] Smallest decisive experiments/probes to reduce uncertainty fast
- [DYNAMICS] Second-order effects, incentives, strategic/adversarial responses
- [STAKEHOLDERS] Users, operators, buyers, regulators; adoption frictions
- [SIGNALS] Leading indicators, telemetry, and field signals to watch
Format:
- Numbered list with 12–16 items
- Begin each item with one of the bracket tags above (e.g., [GAPS]) followed by a sharp question
- One line per item
Surface "unknown unknowns" via analogy and anomaly patterns
When to use: When you suspect hidden variables; use analogies/anomalies to reveal blind regimes.
Propose sharp questions that reveal unknown unknowns using analogies, anomalies, and regime shifts.
Format: Numbered list (10–14). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Probe: [ANALOGY|ANOMALY|SHIFT]
- Reveal: ...
Generate intentionally flawed, paradoxical, or misleading questions to challenge assumptions
When to use: To jolt thinking when the conversation feels stuck or overly confident.
Generate 10 provocative questions that are intentionally flawed, paradoxical, or misleading to challenge assumptions and deepen thinking.
Guidance:
- Be intellectually disruptive: violate common logic or accepted framing
- Be playfully incorrect or oversimplified in ways that invite correction
- Use paradoxes, self-reference, category errors, and impossibilities
- Aim for questions that expose hidden assumptions and blindspots
Generate design optimization questions following natural leverage points
When to use: When improving an existing design; target true leverage points, not cosmetics.
Analyze this design/system and generate the most impactful optimization questions. Focus on where the actual leverage points are, not artificial categories.
Approach:
1. First, identify the 2-3 dimensions where optimization would have the highest impact for THIS specific topic
2. Then generate 5-10 questions total, distributed naturally across those high-impact areas
3. Each question should reveal a specific optimization opportunity or hidden constraint
4. Questions can span multiple concerns (e.g., a performance issue that affects UX)
Quality criteria:
- Density over coverage: 3 excellent questions in one critical area beats 7 mediocre ones across all areas
- Specificity: Questions should be precise to this topic, not portable to any system
- Actionability: Each question should lead to concrete design decisions
- Non-obviousness: Surface hidden constraints and counter-intuitive optimization paths
Format:
- Brief analysis (1-2 sentences) of where the key optimization opportunities lie
- Numbered list of questions (5-10 total)
Enumerate near-neighbor ideas that become viable if a current bet holds
When to use: After a core thesis emerges; identify near-term expansions and compounding bets.
Enumerate adjacent ideas that become unlocked if the current thesis holds. Favor compounding and platform effects.
Format: Numbered list (10–14). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Idea: ...
- Dependency: ...
- Compounding: ...
Combine dimensions to generate non-obvious design variants
When to use: When ideation is converging too fast; recombine dimensions to uncover novel variants.
Generate design variants by combining 3–5 key dimensions; highlight the surprising combos.
Format: Numbered list (10–14). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Combo: ...
- Why: ...
- Tradeoff: ...
Explore max/min designs at the limits of physics/software/human factors
When to use: When exploring boundaries and stress conditions; test designs at extremes.
Explore extreme versions of the idea (max performance, min cost, zero-trust, offline-only, etc.).
Format: Numbered list (10). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets:
- Extreme: ...
- Change: ...
- Risk: ...
Define falsifiable hypotheses and explicit kill criteria/decision thresholds
When to use: When you need go/no-go gates for a bet; make hypotheses testable and decisions crisp.
Propose the minimal set of falsifiable hypotheses that govern this topic, with explicit decision thresholds (kill, pivot, double-down).
For each item include:
- Hypothesis: concise, testable claim
- Evidence: what observation would refute it fast
- Threshold: numeric or categorical boundary for kill/pivot
- Time/Cost: rough order (hours/days) and resources
Constraints:
- Prefer hypotheses that de-risk core mechanisms, not cosmetics.
- Keep each item on one line to stay scannable.
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets for the fields above
Minimum Viable Experiments to reduce top uncertainties quickly
When to use: Right after risks are known; design smallest probes to collapse uncertainty fast.
Design the smallest decisive experiments to reduce the top uncertainties. Favor scrappy probes over polish.
For each item include:
- Goal: uncertainty to reduce
- Method: experiment design
- Signal: what success looks like (leading indicator)
- Time/Cost: hours/days, rough effort
- Risk: failure/ethics considerations
Format: Numbered list (8–12). For each item:
- First line: short handle (≤12 words)
- Then sub-bullets for the fields above
Define a term or concept concisely in research context
When to use: When a term is overloaded or unclear; add a crisp, context-aware definition.
Provide a crisp definition of the term or concept in context.
Style:
- 1–2 sentences max
- Plain language; avoid jargon unless crucial
- Disambiguate briefly if the term is overloaded
Generate thought fragments or poetic contradictions
When to use: When you want evocative sparks over answers; capture tensions and half-formed ideas.
Generate 5–10 thought fragments, impressions, or poetic contradictions that live in the space between questioning and knowing.
Style:
- Do not resolve, explain, or direct
- Offer fragments, intuitions, tensions, metaphors, paradoxes
- Feel incomplete but suggestive—like glimpses of questions not fully formed
Generate multiple conversational responses that continue the dialogue naturally
When to use: When you want several next-turn options with varied tone and angle.
Generate 7 diverse, self-contained responses that continue the conversation naturally.
Guidance:
- Build on the context; deepen or clarify the current idea
- Conversational but informative; 2–4 sentences each
- Offer distinct perspectives, tones, or angles
- End with a statement, not a question; no follow-up questions
Generate a concise, insightful reply that continues the conversation
When to use: Default next-turn reply; keep it short, insightful, and forward-moving.
Respond to the message concisely. Aim to either clarify, offer a useful explanation/insight, or invite further exploration.
Generate a condensed summary that captures the core ideas
When to use: When a branch is long; synthesize children into a concise, faithful parent summary.
Create a synthesis that intelligently combines multiple perspectives.
First, recognize what type of responses you're synthesizing:
- Multiple critiques → Create a meta-critique highlighting the most significant issues
- Multiple questions → Synthesize into the most penetrating composite questions
- Multiple ideas → Combine into the strongest unified concept
- Multiple analyses → Merge insights into comprehensive understanding
- Mixed responses → Extract the throughline that connects them
Principles:
- Act at the same semantic level as the inputs (critique of critiques, question about questions, etc.)
- Don't "fix" or "respond to" the inputs - synthesize them
- Extract patterns and recurring themes
- Combine complementary insights, resolve conflicts by choosing stronger arguments
- Include specific examples, numbers, evidence from the best inputs
- Aim for 60-80% compression
Present the synthesis as THE definitive response. The result should feel like the smartest person in the room who heard all perspectives and is now giving THE answer.
-
Paste after your topic: Write your topic/question first, then paste the relevant prompt after it.
-
Combine prompts: You can run multiple prompts in sequence to build up analysis (e.g., Assumptions Audit → Kill Criteria → MVP Experiments).
-
Adapt the format: If you need fewer or more items, adjust the numbers in the format section.
-
Follow up: Use the output as input for the next prompt to go deeper.