| name | creative-problem-solver |
| description | Lateral-thinking playbook for complex solution design; use for architecture/migration/re-architecture, multi-system integration, high-uncertainty tradeoffs, conflict resolution, or when you need options, reframes, and a Quick Win / Strategic Play / Transformative Move. |
Creative Problem Solver
When to use
- Progress is stalled or blocked π§.
- Repeated attempts fail in the same way π.
- The user asks for new angles, options, or "outside the box" ideas π‘.
- The problem is complex, tricky, tangled, or multi-constraint π§©.
- The problem is novel, ambiguous, high-uncertainty, or cross-domain πͺοΈ.
- The team is circling the same constraints without new levers π§±.
Round contract (one turn per round)
- Deliver options, then stop and ask for human input before executing.
- Update the Knowledge Snapshot with new signals, constraints, and decisions.
- Record a Decision Log entry and Assumptions/Constraints list.
Quick start
- State why the current tactic fails (one sentence).
- Reframe the constraint (use 1 technique from the Reframing toolkit).
- Pick 1β3 techniques from the Library to generate options.
- Propose the portfolio: Quick Win β‘, Strategic Play π―, Transformative Move π.
- For each option, include: expected signal and escape hatch.
- Score options (1β5) for Signal, Cost, Reversibility, Time.
- Log decisions + assumptions/constraints.
- Ask the user to choose or adjust constraints.
- Close with an Insights Summary.
Mode check
- Pragmatic Mode β (default): ship-this-week options only.
- Visionary Mode π: only when the user asks for long-term strategy or systemic change.
Reframing toolkit
- Inversion: do the opposite of the current approach.
- Analogy transfer: map to a solved domain and borrow the pattern.
- Constraint extremes: set a key variable to zero or infinity.
- First principles: rebuild from basic facts, not inherited assumptions.
Portfolio rule
Every response must include:
- Quick Win β‘: smallest experiment that changes the information you have.
- Strategic Play π―: medium effort, leverages a structural advantage.
- Transformative Move π: changes the system shape or assumptions.
Option template
Quick Win β‘:
- Expected signal:
- Escape hatch:
Strategic Play π―:
- Expected signal:
- Escape hatch:
Transformative Move π:
- Expected signal:
- Escape hatch:
Scoring rubric (simple 1β5, no weights)
Purpose: compare options quickly, not to compute a "correct" answer.
- Signal: how much new information this yields (1=low, 5=high).
- Cost: effort/complexity to try (1=low, 5=high).
- Reversibility: ease of undoing (1=hard, 5=easy).
- Time: elapsed time to learn (1=fast, 5=slow).
Prefer: high Signal, high Reversibility, low Cost, low Time.
Ideation β prioritization workflow
- Diverge: generate 10β30 ideas (no judging).
- Cluster: group by theme or lever.
- Score: apply the rubric.
- Pick top 3 (one per portfolio tier).
Decision Log (fill every round)
- Decision:
- Rationale:
- Alternatives considered:
- Evidence / signal:
- Reversible? (Y/N):
- Next decision point:
Assumptions & Constraints
- Assumptions to validate:
- Known constraints (time, budget, policy, tech):
- Unknowns to de-risk:
- Non-goals / out of scope:
Knowledge Snapshot (update every round)
- New facts learned:
- New risks / constraints:
- Signals that changed the plan:
- Open questions:
Technique picker (fast selection)
- Need to mutate an existing thing β SCAMPER.
- Need lots of ideas fast β Brainwriting 6-3-5 or Brainstorming.
- Need structured combinations β Morphological Analysis.
- Need to resolve contradictions β TRIZ.
- Need parallel perspectives β Six Thinking Hats.
- Need user-centric clarity β Design Thinking + Storyboard.
- Need reframes β Lateral Thinking / Random Stimulus / Forced Connections.
Technique Library (use 1β3 per round)
Creative Problem Solving (CPS) Cycle π§
- Purpose: move from messy challenge to tested solution.
- How: Clarify β Ideate β Develop β Implement.
- Example: "Reduce churn" β clarify drivers, ideate 30 retention ideas, develop top 3 tests, implement one pilot.
Brainstorming (Osborn rules) π
- Purpose: flood the space with ideas.
- How: defer judgment, go for quantity, build on others, welcome wild ideas.
- Example: "Cut build time" β 20 wild ideas in 10 minutes before evaluating.
Brainwriting 6-3-5 βοΈ
- Purpose: generate lots of ideas quietly and quickly.
- How: 6 people, 3 ideas each, 5 rounds; pass ideas along.
- Example: "Improve onboarding" β 18β30 ideas without groupthink.
SCAMPER π§
- Purpose: systematically transform an existing thing.
- How: Substitute, Combine, Adapt, Modify, Put to another use, Eliminate, Reverse.
- Example: "Faster code reviews" β Eliminate long templates, Combine review+test, Reverse by batching reviews.
Six Thinking Hats π©
- Purpose: force parallel viewpoints.
- How: cycle hats (facts, emotions, risks, benefits, creativity, process).
- Example: "Adopt new framework" β run hats to separate risks from gut feel.
TRIZ (Contradictions + 40 Principles) π§ͺ
- Purpose: resolve tradeoffs without compromise.
- How: express contradiction, pick principles that dissolve it.
- Example: "Increase performance without memory use" β apply "Segmentation" or "Dynamicity" to scope caches.
Morphological Analysis π§±
- Purpose: explore combinations across dimensions.
- How: define key parameters, list options for each, mix-and-match.
- Example: "New CLI" β axes: input, output, auth, mode; combine to surface novel designs.
Synectics π
- Purpose: use analogies to break patterns.
- How: direct analogy β personal analogy β symbolic analogy.
- Example: "Queue system" β think of a grocery line, then a beehive, then "breathing" to smooth surges.
Lateral Thinking / Provocation (PO) β‘
- Purpose: break linear assumptions with provocative moves.
- How: state a provocative idea, then extract a practical insight.
- Example: "No login at all" β learn that first-time use should be zero-friction.
Random Stimulus / Random Word π²
- Purpose: force unexpected associations.
- How: pull a random word/image, map its properties to the problem.
- Example: Random word "bridge" β "bridge" as a guided handoff between steps.
Forced Connections π
- Purpose: connect unrelated concepts to spark ideas.
- How: pair two random items and ask "how could these connect?"
- Example: "Checkout flow + airport security" β pre-checklist, fast lanes, clear signage.
Reverse Brainstorming π
- Purpose: find failure modes, then invert them.
- How: ask "How do we make this worse?" then reverse answers.
- Example: "Make support slower" β identify delays, then remove them.
Mind Mapping πΊοΈ
- Purpose: expand and organize idea space.
- How: central node β branches β sub-branches.
- Example: "Observability revamp" β branches for logs/metrics/traces/alerts.
Affinity Diagramming π§©
- Purpose: discover themes across many ideas.
- How: cluster notes without labels, then name clusters.
- Example: "User feedback" β cluster into onboarding, pricing, reliability.
How Might We (HMW) β
- Purpose: reframe problems into solvable questions.
- How: convert a complaint into an HMW question.
- Example: "Deploys are scary" β "How might we make deploys feel reversible?"
Design Thinking π§
- Purpose: build user-centric solutions.
- How: Empathize β Define β Ideate β Prototype β Test.
- Example: "New admin UI" β shadow admins, define pain, prototype, test.
Crazy 8s / Speedy Eights βοΈ
- Purpose: produce rapid variations fast.
- How: 8 sketches in 8 minutes (or 4β8 minutes for 4β8 sketches).
- Example: "Dashboard layout" β 8 distinct layouts before discussing.
Storyboarding ποΈ
- Purpose: visualize end-to-end flow.
- How: sketch key frames of the user journey.
- Example: "Signup β first success" in 6 frames to spot friction.
Lotus Blossom πΈ
- Purpose: expand ideas from a core problem outward.
- How: center problem, fill 8 petals with ideas, then expand each petal.
- Example: "Reduce incident MTTR" β petals for tooling, process, training, automation.
Deliverable format (response)
- Why current tactic fails (1 sentence).
- Reframe used + new lever.
- Portfolio options with expected signals and escape hatches.
- Scorecard (1β5) and quick rationale.
- Decision Log + Assumptions/Constraints.
- Human Input Required: choose option(s) or update constraints.
- Insights Summary with next actions.
Activation cues
- "need options" / "give me options"
- "stuck" / "blocked" / "nothing works"
- "outside the box" / "new ideas" / "fresh angles"
- "complex" / "tricky" / "messy" / "tangled"
- "ambiguous" / "uncertain" / "unknowns"
- "novel" / "never done this"
- "too many constraints"
- "conflicting requirements"
- "architecture" / "system design" / "re-architecture"
- "migration" / "cutover" / "zero-downtime"
- "multi-system integration" / "cross-service"
- "offline sync" / "conflict resolution"
- "event sourcing" / "CQRS"
- "plugin system" / "sandboxing"
- "sharding" / "rebalancing"
- "failover" / "disaster recovery"
- "high-availability" / "multi-region"
- "feature flag migration" / "schema evolution"
Examples:
- "Design multi-tenant RBAC/ABAC with audit trails across services"
- "Plan a zero-downtime monolith-to-services migration with phased cutover"
- "Build offline-first sync with conflict resolution for web + mobile"
- "Re-architect CI to cut time 70% across polyrepo + mixed languages"
- "Introduce event sourcing + CQRS without breaking existing APIs"
- "Design a plugin system with sandboxing + versioned APIs"
- "Shard the datastore with consistent hashing and safe rebalancing"
- "Migrate auth providers with dual-write + rollback strategy"
- "Implement feature-flagged schema evolution at scale"
- "Design a multi-region failover strategy with tight RTO/RPO"