See every problem from every angle.
+-------------------------------------------+
| T H E C U B E |
| Multi-Dimensional Problem Solver |
| v1.0 | 30 Positions |
+-------------------------------------------+
The Cube is a problem-solving skill for Claude Code that maps six cognitive frameworks to the six faces of a Rubik's cube. Your problem goes at the core. The cube rotates through every geometric combination -- faces, edges, axes, corners -- producing 30 distinct analysis points before returning to home position with a grand synthesis.
The throughline: every rotation forces you out of the frame the problem arrived in. The problem's native framing is usually part of what's trapping you.
git clone https://github.com/mrdulasolutions/TheCube.git
cd TheCube
./install.sh/cube My startup has 6 months of runway left and our main product isn't getting traction
That's it. The Cube will crystallize your problem, rotate through all 30 positions, and produce a structured report with ranked solutions and simulation scenarios.
flowchart LR
P["YOUR PROBLEM"] --> C{"THE CUBE"}
C --> F1["Phase 1\nFace Scan\n6 Positions"]
F1 --> F2["Phase 2\nEdge Rotations\n12 Positions"]
F2 --> F3["Phase 3\nAxis Tensions\n3 Positions"]
F3 --> F4["Phase 4\nCorner Rotations\n8 Positions"]
F4 --> F5["Phase 5\nGrand Synthesis\n1 Position"]
F5 --> R["RANKED SOLUTIONS\n+ SIMULATIONS"]
style P fill:#ff6b6b,color:#fff,stroke:none
style C fill:#1a1a2e,color:#fff,stroke:none
style F1 fill:#16213e,color:#fff,stroke:none
style F2 fill:#0f3460,color:#fff,stroke:none
style F3 fill:#533483,color:#fff,stroke:none
style F4 fill:#e94560,color:#fff,stroke:none
style F5 fill:#1a1a2e,color:#fff,stroke:none
style R fill:#00b894,color:#fff,stroke:none
The Cube ships with four domain-specific stacks -- four complete Rubik's cubes, each with six frameworks tailored to a specific discipline. Same rotation protocol, different lenses.
flowchart TB
subgraph CORE["CORE /cube"]
direction LR
C1[Inversion] ~~~ C2[First Principles]
C3[Steelmanning] ~~~ C4[Second-Order]
C5[Pre-Mortem] ~~~ C6[Boundary]
end
subgraph SALES["SALES /cube-sales"]
direction LR
S1[Klaff] ~~~ S2[Rackham]
S3[Carnegie] ~~~ S4[Dixon]
S5[Ziglar] ~~~ S6[Sandler]
end
subgraph MARKETING["MARKETING /cube-marketing"]
direction LR
M1[Christensen] ~~~ M2[Ries & Trout]
M3[Miller] ~~~ M4[Kim & Mauborgne]
M5[Moore] ~~~ M6[Chen/Reforge]
end
subgraph CODING["CODING /cube-coding"]
direction LR
D1[Decomposition] ~~~ D2[Root Cause]
D3[Constraints] ~~~ D4[Trace]
D5[Edge Cases] ~~~ D6[MVP Proof]
end
style CORE fill:#1a1a2e,color:#fff,stroke:#e94560
style SALES fill:#1a1a2e,color:#fff,stroke:#ff6b6b
style MARKETING fill:#1a1a2e,color:#fff,stroke:#533483
style CODING fill:#1a1a2e,color:#fff,stroke:#00b894
| Stack | Command | Frameworks |
|---|---|---|
| Core | /cube |
Inversion, First Principles, Steelmanning, Second-Order, Pre-Mortem, Boundary |
| Sales | /cube-sales |
Frame Control (Klaff), SPIN (Rackham), Rapport (Carnegie), Challenger (Dixon), Need Satisfaction (Ziglar), Pain Qualification (Sandler) |
| Marketing | /cube-marketing |
Jobs to Be Done (Christensen), Positioning (Ries & Trout), StoryBrand (Miller), Blue Ocean (Kim & Mauborgne), Crossing the Chasm (Moore), Growth Loops (Chen/Reforge) |
| Coding | /cube-coding |
Decomposition, Root Cause Analysis, Constraint Mapping, Trace the Execution, Edge Case Analysis, Minimum Viable Proof |
Run /cube-stack to see the full directory with framework descriptions and recommendations.
Six elite sales methodologies that create three productive tensions:
- Control vs Discovery (Klaff vs Rackham) -- Who leads the conversation?
- Desire vs Pain (Ziglar vs Sandler) -- Toward pleasure or away from pain?
- Warmth vs Hard Truth (Carnegie vs Challenger) -- Be liked or be respected?
Named rotations include: The Pitch, The Power Play, The Velvet Hammer, The Pain Funnel, The Trusted Advisor, The Charmer, The Surgeon, The Prosecutor, and more.
Six foundational marketing frameworks that create three productive tensions:
- Truth vs Perception (JTBD vs Positioning) -- What they need vs what they believe
- Adoption vs Growth (Chasm vs Loops) -- What blocks growth vs what compounds it
- Clarify vs Reinvent (StoryBrand vs Blue Ocean) -- Perfect the message vs rewrite the game
Named rotations include: The Unmet Job, The Beachhead, The Category Claim, The Flywheel, The Launch Story, The Rocketship, The Category King, and more.
Six engineering problem-solving frameworks that create three productive tensions:
- Breadth vs Depth (Decomposition vs Root Cause) -- Break it apart vs dig to the bottom
- Breaks vs Works (Edge Cases vs MVP Proof) -- What fails vs what succeeds
- Theory vs Reality (Constraints vs Trace) -- What should happen vs what does happen
Named rotations include: The Fragmentation Test, The Debugger, The Hidden Wall, The Fault Line, The Architecture Spike, The Bug Hunt, The Surgical Fix, and more.
| Command | Positions | Best For |
|---|---|---|
/cube [problem] |
30 | Full analysis (Core stack) |
/cube-sales [problem] |
30 | Sales deal strategy and methodology |
/cube-marketing [problem] |
30 | Marketing strategy and growth |
/cube-coding [problem] |
30 | Debugging, architecture, technical problems |
/cube-quick [problem] |
10 | Fast scan (any stack with --sales, --marketing, --coding) |
/cube-face [framework] [problem] |
1 (deep) | Deep dive into any framework from any stack |
/cube-guided [problem] |
30 | Interactive mode (any stack) |
/cube-stack |
-- | List all stacks and frameworks |
/cube-feedback |
-- | Rate an analysis to improve the tool |
The complete 30-position analysis for the chosen stack. Five phases, every geometric combination, full synthesis with solution ranking and simulation scenarios.
A condensed 10-position analysis: all 6 face scans, the 3 axis tensions, and a focused synthesis. Supports all stacks:
/cube-quick My startup is running out of runway
/cube-quick --sales This enterprise deal has stalled at procurement
/cube-quick --marketing We are getting traffic but no conversions
/cube-quick --coding The API response times have doubled since the last deploy
Applies ONE framework from any stack with maximum depth. Full 4-phase deep dive protocol:
/cube-face inversion Why are we losing customers?
/cube-face klaff My pitch keeps falling flat with VPs
/cube-face jtbd Our users churn after 30 days
/cube-face root-cause The payment service crashes every Monday morning
Same 30 positions as the full rotation, but pauses after each phase. Supports all stacks with --sales, --marketing, --coding flags.
Structured feedback collection after an analysis. Saves to .cube/feedback/ for telemetry.
graph TB
subgraph CUBE["THE CUBE — Face Layout"]
direction TB
U["UP (U)\nPre-Mortem"]
L["LEFT (L)\nSteelmanning"]
F["FRONT (F)\nInversion"]
R["RIGHT (R)\nSecond-Order"]
B["BACK (B)\nFirst Principles"]
D["DOWN (D)\nBoundary"]
U --- F
L --- F
F --- R
F --- D
F -.-> B
end
style U fill:#533483,color:#fff,stroke:none
style L fill:#0f3460,color:#fff,stroke:none
style F fill:#e94560,color:#fff,stroke:none
style R fill:#ff6b6b,color:#fff,stroke:none
style B fill:#16213e,color:#fff,stroke:none
style D fill:#00b894,color:#fff,stroke:none
style CUBE fill:#0a0a1a,color:#fff,stroke:#333
| Face | Framework | Core Question |
|---|---|---|
| F | Inversion | How would I guarantee this gets worse? |
| B | First Principles | What's actually true when I strip all assumptions? |
| L | Steelmanning | What's the strongest case for NOT solving this? |
| R | Second-Order Thinking | What new problems does my solution create? |
| U | Pre-Mortem | It already failed. What killed it? |
| D | Boundary Conditions | What happens at zero? At infinity? At 1000x? |
They are orthogonal (each sees something the others cannot), adversarial to framing (none accept the problem as presented), universally applicable (they work on technical, business, personal, and creative problems), and they compound in combination (the edges, axes, and corners produce insights no single framework could).
Read ETHOS.md for the full philosophy.
The cube arranges frameworks so opposite faces create productive tension:
graph LR
subgraph X-AXIS["X-AXIS: Destruction vs Construction"]
I["Inversion\n(F)"] <-->|"tension"| FP["First Principles\n(B)"]
end
subgraph Y-AXIS["Y-AXIS: Temporal vs Structural Failure"]
PM["Pre-Mortem\n(U)"] <-->|"tension"| BC["Boundary\n(D)"]
end
subgraph Z-AXIS["Z-AXIS: Resistance vs Consequences"]
SM["Steelmanning\n(L)"] <-->|"tension"| SO["Second-Order\n(R)"]
end
style I fill:#e94560,color:#fff,stroke:none
style FP fill:#16213e,color:#fff,stroke:none
style PM fill:#533483,color:#fff,stroke:none
style BC fill:#00b894,color:#fff,stroke:none
style SM fill:#0f3460,color:#fff,stroke:none
style SO fill:#ff6b6b,color:#fff,stroke:none
style X-AXIS fill:#0a0a1a,color:#fff,stroke:#e94560
style Y-AXIS fill:#0a0a1a,color:#fff,stroke:#533483
style Z-AXIS fill:#0a0a1a,color:#fff,stroke:#0f3460
| Axis | Faces | Tension |
|---|---|---|
| X | Inversion vs First Principles | Destruction vs Construction |
| Y | Pre-Mortem vs Boundary | Temporal Failure vs Structural Failure |
| Z | Steelmanning vs Second-Order | Resistance vs Consequences |
graph LR
subgraph P1["PHASE 1"]
F1["Face Scan\n6 solo analyses"]
end
subgraph P2["PHASE 2"]
F2["Edge Rotations\n12 adjacent pairs"]
end
subgraph P3["PHASE 3"]
F3["Axis Tensions\n3 opposite-pair\ndialectics"]
end
subgraph P4["PHASE 4"]
F4["Corner Rotations\n8 triple-framework\ntriangulations"]
end
subgraph P5["PHASE 5"]
F5["Grand Synthesis\nRanked solutions\n+ simulations"]
end
P1 --> P2 --> P3 --> P4 --> P5
style P1 fill:#16213e,color:#fff,stroke:#e94560
style P2 fill:#0f3460,color:#fff,stroke:#e94560
style P3 fill:#533483,color:#fff,stroke:#e94560
style P4 fill:#e94560,color:#fff,stroke:#e94560
style P5 fill:#1a1a2e,color:#fff,stroke:#00b894
style F1 fill:transparent,color:#fff,stroke:none
style F2 fill:transparent,color:#fff,stroke:none
style F3 fill:transparent,color:#fff,stroke:none
style F4 fill:transparent,color:#fff,stroke:none
style F5 fill:transparent,color:#fff,stroke:none
Each framework applied independently. Six solo analyses, each ending with a Key Insight -- a single-sentence distillation. After all six, a Complexity Assessment rates the problem's depth, risk density, cascade potential, and boundary sensitivity.
Every pair of adjacent frameworks combined. Twelve cross-analyses, each named for its character:
| Position | Name | Combination | What It Reveals |
|---|---|---|---|
| 7 | The Autopsy | Inversion + Pre-Mortem | Overlap between deliberate and actual failure |
| 8 | The Cascade | Inversion + Second-Order | Chain reactions from worst-case scenarios |
| 9 | The Stress Test | Inversion + Boundary | What breaks first under maximum adversity |
| 10 | Devil's Advocate | Inversion + Steelmanning | The case for making it worse on purpose |
| 11 | Foundation Check | First Principles + Pre-Mortem | Which foundational truth is the real failure point |
| 12 | The Projection | First Principles + Second-Order | Downstream effects hidden by assumptions |
| 13 | Bedrock Test | First Principles + Boundary | Which "fundamentals" are actually contextual |
| 14 | The Challenge | First Principles + Steelmanning | First principles that can be defeated |
| 15 | The Fortress | Pre-Mortem + Steelmanning | Where "it's dead" and "don't try" reinforce each other |
| 16 | The Domino Map | Pre-Mortem + Second-Order | Failure modes with the worst cascades |
| 17 | The Pressure Test | Boundary + Steelmanning | Whether the case against solving holds at extremes |
| 18 | The Horizon Scan | Boundary + Second-Order | Effects that only emerge at scale |
Opposite faces create the deepest tensions -- thesis vs antithesis:
| Position | Name | Tension |
|---|---|---|
| 19 | The Mirror | Inversion x First Principles |
| 20 | The Crucible | Pre-Mortem x Boundary |
| 21 | The Paradox | Steelmanning x Second-Order |
Three frameworks meet at each corner. Each corner selects exactly one from each opposite pair (one from F/B, one from U/D, one from L/R), ensuring maximum dimensional coverage:
| Position | Name | Frameworks | What Survives |
|---|---|---|---|
| 22 | The Gauntlet | Inversion + Pre-Mortem + Steelmanning | Triple adversity -- what still matters? |
| 23 | The Avalanche | Inversion + Pre-Mortem + Second-Order | Total failure cascade mapped |
| 24 | The Siege | Inversion + Boundary + Steelmanning | Worth doing even at maximum adversity |
| 25 | The Labyrinth | Inversion + Boundary + Second-Order | Where complexity simplifies under stress |
| 26 | The Tribunal | First Principles + Pre-Mortem + Steelmanning | Judgment day -- who is right? |
| 27 | The Oracle | First Principles + Pre-Mortem + Second-Order | What should have been foreseen |
| 28 | The Anvil | First Principles + Boundary + Steelmanning | What is genuinely unbreakable |
| 29 | The Telescope | First Principles + Boundary + Second-Order | The long view -- what dominates at scale |
Grand synthesis:
- Executive Summary -- 4-line decision-maker brief
- Pattern Detection -- recurring themes across rotations
- Convergence Map -- where multiple rotations independently agree (highest-confidence findings)
- Divergence Report -- unresolved tensions (often where the real decision lives)
- Blind Spot Report -- insights visible from only one angle
- Solution Ranking -- candidates scored by robustness, survivability, and anti-fragility
- Simulation Scenarios -- agent-testable setups for each solution path
- Final Verdict -- the clearest path forward
+-------------------------------------------+
| T H E C U B E |
| Multi-Dimensional Problem Solver |
| v1.0 | 30 Positions |
+-------------------------------------------+
# THE CUBE -- Analysis Report
## Problem: Startup runway crisis with stalled product traction
### Rotation Date: 2026-04-13
### Total Analysis Points: 30
## THE CORE
**Problem:** We have 6 months of runway remaining and our core product
has plateaued at 200 MAU despite 18 months of development.
**Stakes:** Company survival. Team of 8 loses jobs. Investors lose capital.
**Current Frame:** "We need more features to attract users."
**Domain:** Business/Strategic
**Time Sensitivity:** Immediate
--- PHASE 1: FACE SCAN --- Positions 1-6 --- Progress 0/30
### [F] Position 1: Inversion
To guarantee failure: keep building features nobody asked for, avoid
talking to the 200 users who DID sign up, burn runway on paid acquisition
before finding product-market fit, and refuse to consider a pivot...
**Key Insight:** The fastest way to die is to keep building without
learning why the 200 existing users stayed.
[...28 more positions...]
### EXECUTIVE SUMMARY
**The real problem:** This is not a feature gap -- it is a learning gap.
**The clearest path:** Interview the 200 existing users this week.
**The biggest risk:** Running out of runway while still guessing.
**The key trade-off:** Speed of learning vs. completeness of product.
The geometry is not decorative:
pie title 30 Analysis Positions
"Faces (6)" : 6
"Edges (12)" : 12
"Axes (3)" : 3
"Corners (8)" : 8
"Synthesis (1)" : 1
| Element | Count | What It Represents |
|---|---|---|
| Faces | 6 | Independent viewpoints |
| Edges | 12 | Natural adjacencies where viewpoints meet |
| Axes | 3 | Fundamental tensions between opposing viewpoints |
| Corners | 8 | Triangulation points where three viewpoints converge |
| Core | 1 | The problem itself, which never moves |
Every element maps to a real geometric relationship. The cube completes all rotations and returns to home position -- every piece is back where it started, but you've seen the problem from every possible combination of angles.
One command. Works everywhere.
git clone https://github.com/mrdulasolutions/TheCube.git
cd TheCube
./install.shThis installs all 9 skills to ~/.claude/skills/ in both standalone and directory formats, so /cube works in Claude Code, Cowork, and Claude Desktop.
Cowork users: Run this in your Mac's Terminal, then start a new Cowork session. The skills will be available immediately.
Manual alternative:
mkdir -p ~/.claude/skills
cp .claude/skills/cube*.md ~/.claude/skills/
for f in .claude/skills/cube*.md; do
name=$(basename "$f" .md)
mkdir -p ~/.claude/skills/$name
cp "$f" ~/.claude/skills/$name/SKILL.md
doneNo install needed — works through Projects:
- Create a new Project on claude.ai
- Upload the skill files from
.claude/skills/as knowledge (start withcube.mdandcube-quick.md) - Paste the contents of
chat/project-instructions.mdinto the project's custom instructions - Start a conversation and say: "Cube this: [your problem]"
See chat/SETUP.md for the full setup guide.
| Action | Claude Code / Cowork | Claude.ai Chat |
|---|---|---|
| Core full rotation | /cube [problem] |
"Cube this: [problem]" |
| Sales stack | /cube-sales [problem] |
"Cube sales: [problem]" |
| Marketing stack | /cube-marketing [problem] |
"Cube marketing: [problem]" |
| Coding stack | /cube-coding [problem] |
"Cube coding: [problem]" |
| Quick scan | /cube-quick [problem] |
"Quick cube: [problem]" |
| Deep dive | /cube-face inversion [problem] |
"Deep dive inversion on [problem]" |
| Guided mode | /cube-guided [problem] |
"Guided cube: [problem]" |
| Stack list | /cube-stack |
"What cubes are there?" |
| Feedback | /cube-feedback |
"Cube feedback" |
The Cube collects no data automatically. All telemetry is opt-in via /cube-feedback.
When you run /cube-feedback after an analysis, it saves structured feedback to .cube/feedback/ in your local project. This data never leaves your machine unless you choose to share it.
To contribute your feedback to improve The Cube:
- Open a GitHub issue with the
feedbacklabel - Include your feedback file (redact sensitive problem details)
- See CONTRIBUTING.md for details
TheCube/
├── .claude/skills/
│ ├── cube.md Core stack (30 positions)
│ ├── cube-sales.md Sales stack (30 positions)
│ ├── cube-marketing.md Marketing stack (30 positions)
│ ├── cube-coding.md Coding stack (30 positions)
│ ├── cube-stack.md Stack directory
│ ├── cube-quick.md Quick scan (10 positions, all stacks)
│ ├── cube-face.md Single framework deep dive (all stacks)
│ ├── cube-guided.md Interactive mode (all stacks)
│ └── cube-feedback.md Feedback and telemetry
├── chat/
│ ├── SETUP.md Claude.ai Projects setup guide
│ └── project-instructions.md Custom instructions for Claude.ai
├── CLAUDE.md Project instructions for Claude Code
├── CONTRIBUTING.md How to contribute
├── ETHOS.md Philosophy and design principles
├── LICENSE MIT
├── README.md This file
└── install.sh One-command global installer (all 9 skills)
Don't ask how to solve it -- ask how to guarantee the problem gets worse. What you're trying to avoid becomes clearer than what you're trying to achieve.
"All I want to know is where I'm going to die, so I'll never go there." -- Charlie Munger
Strip every assumption until you hit something you can verify from scratch. Most "problems" are actually inherited constraints that no longer apply.
Before you solve a problem, build the strongest possible case for why it can't or shouldn't be solved. If you can't defeat the steel man, you don't understand the problem yet.
Every solution creates new problems. Map 2-3 levels deep: "this fixes X, which causes Y, which means Z." Most bad decisions look good at the first-order level.
Project is already dead. What killed it? Forces specificity -- you stop saying "risk" and start naming actual failure modes. Far more honest than forward-looking planning.
Push the problem to its extremes: zero resources, unlimited resources, 1000x scale, one day to solve it. Assumptions that feel load-bearing in the normal range often collapse at the edges -- and that tells you what the problem is actually made of.
Read ETHOS.md for the full philosophy behind The Cube: why these six frameworks, why the cube geometry, what The Cube is and isn't.
Read CONTRIBUTING.md for how to submit feedback, propose improvements, suggest new frameworks, or contribute code.
MIT -- see LICENSE.