A Claude Code-powered Shell-style scenario planning system with multi-specialist consultation, a dedicated impact analysis layer, worldview integration, and rigorous documentation enforcement.
This is a structured scenario planning tool based on the Shell methodology, featuring:
- Dr. Michelle Wells (Moderator) - Facilitates a worldview-first scenario workflow with a dedicated impact translation layer
- Worldview Elicitation - Understands how you think about the topic before exploring scenarios
- 6 Domain Specialist Consultants - Diverse expert perspectives (ecology, geopolitics, culture, technology, economics, contrarian)
- Impact Translation Cast - Named composite personas that translate external scenarios into actor-relative consequences
- Dr. Anya Petrov (Research Specialist) - Multi-source research and fact-checking
- Dr. Mei Chen (Quality Analyst) - Automated quality gates prevent regression to generic analysis
- Worldview Integration - Connects scenarios back to your mental model
- Automated Quality Enforcement - Hooks trigger quality gates, ensure intellectual integrity
┌─────────────────────────────────────────────────────────────────┐
│ PHASE 0: WORLDVIEW ELICITATION │
│ Understand how you think about this topic │
│ OUTPUT: worldview_model.md │
└─────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ PHASES 1-5: EXTERNAL SCENARIO PLANNING │
│ Focal question → Predetermined → Uncertainties → Scenarios │
│ OUTPUT: 4 divergent scenarios │
└─────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ PHASE 6A / 6B: IMPACT → STRATEGY │
│ Translate scenarios into actor-relative consequences, then │
│ test preparations, positioning, and response options │
│ OUTPUT: impact_analysis.md, strategy_analysis.md │
└─────────────────────────────────────────────────────────────────┘
│
▼
┌─────────────────────────────────────────────────────────────────┐
│ PHASE 7: WORLDVIEW INTEGRATION │
│ Connect scenarios, impacts, and responses to your lens │
│ OUTPUT: worldview_integration.md │
└─────────────────────────────────────────────────────────────────┘
Required: Perplexity CLI (pp-cli)
npm install -g @dbmcco/pp-cli
pp --versionRequired: worldview-elicitor skill
The worldview-elicitor skill must be available in your Claude Code environment. See Skills Setup below.
git clone https://github.com/dbmcco/shell-scenario-panel.git
cd shell-scenario-panel
# Start Claude Code or Codex conversationJust describe your scenario planning needs:
"I need to develop scenarios for the future of AI regulation."
or
"Let's explore future scenarios for renewable energy infrastructure."
She will:
- Initialize the scenario automatically
- Elicit your worldview - understand how you think about this topic
- Guide you through the full workflow
- Consult specialists as needed
- Translate scenarios into impact - what the external futures mean for the relevant actor
- Test strategies or responses - preparedness, positioning, and other decisions
- Integrate scenarios with your worldview - connect findings to your mental model
- Ensure quality and documentation
You never need to run scripts manually. Dr. Wells manages the technical details.
Use this section if you are operating as the facilitator agent inside this repo.
# Claude sessions
.claude/session-start.sh
# Codex sessions
.codex/session-start.shUse --scenario SCENARIO-YYYY-NNN, --new, or --monitor SCENARIO-YYYY-NNN as needed.
- Check
resources/for user-provided files (ignoreREADME.md,.gitkeep,.DS_Store) - Ask whether to scan and incorporate those materials
- If yes, run
.claude/lib/resources-intake.sh "$SCENARIO_ID" - Review
scenarios/active/$SCENARIO_ID/phase_0_discovery/materials_index.mdwith the user before proceeding
- Treat
prompts/moderator.mdas the canonical facilitation sequence - Complete Phase 0 deliverables before external analysis:
worldview_model.mdphase_0_discovery/internal_baseline.mdphase_0_discovery/context_packet.md- Do not bypass transcript checks, metadata updates, or quality checkpoints
- Claude agent instructions:
CLAUDE.md - Codex agent instructions:
CODEX.md
If there is any conflict, follow the agent-specific file for your runtime.
Place materials in resources/ before starting. The panel performs best when grounded in concrete context about your focal topic.
| Document | Why It Matters |
|---|---|
| Focal Topic Brief | 1-2 pages defining what you want to explore, including time horizon (5, 10, 20 years) and geographic/industry scope |
| Current State Snapshot | What the world looks like today in your focal area — statistics, trends, key players, recent developments |
| Document | Why It Matters |
|---|---|
| Historical Trend Data | How your focal area has evolved over the past 5-20 years; patterns reveal what's truly predetermined vs uncertain |
| Stakeholder Map | Key actors (companies, governments, institutions) and their interests; Marcus (geopolitics) and Sarah (economics) need this |
| Explicit Constraints | What you already know is locked in — budgets, infrastructure timelines, regulatory frameworks, demographic shifts |
| Prior Analysis | Existing strategy documents, forecasts, or expert opinions you've gathered; specialists can build on or challenge these |
| Document | Why It Matters |
|---|---|
| Technology Roadmaps | Capability timelines for relevant technologies; Kenji (futurist) uses these to identify inflection points |
| Cultural Context | How different communities think about your topic; Aisha (anthropologist) grounds scenarios in lived experience |
| Contrarian Views | Perspectives that challenge conventional wisdom; Jamie (contrarian) amplifies these to stress-test scenarios |
- Define your time horizon explicitly — scenarios for 2030 require different inputs than 2050
- Include sources and dates — specialists weight recent data more heavily
- Separate facts from assumptions — label what you know vs what you believe
- Multiple perspectives welcome — conflicting expert views often reveal critical uncertainties
Purpose: Understand how you think about the topic before exploring external scenarios.
Process:
- Dr. Wells invokes the worldview-elicitor skill
- Natural conversation to surface your beliefs, reasoning, and uncertainties
- Captures what would change your mind (cruxes)
- Documents your mental models and frameworks
Output: worldview_model.md
Also captured: phase_0_discovery/internal_baseline.md (structured base case and risk posture)
Why this matters:
- Scenarios become more relevant (we know what lens you're looking through)
- Helps identify blind spots (where your worldview differs from scenario implications)
- Creates better buy-in (scenarios framed in your language/concepts)
- Reveals which scenarios you'll naturally resist or accept
Clarify decision context, time horizon, scope, and assumptions
Map trends already in motion (demographics, infrastructure, debt, climate)
Surface factors that could go multiple ways and matter for decisions
Create 4 plausible, divergent future scenarios with memorable names
Define observable indicators for each scenario
Translate external scenarios into actor-relative consequences using the impact translation cast
Explore strategy or response performance across scenarios using impact_analysis.md as mandatory input
Purpose: Connect scenarios back to your mental model and prepare you for multiple futures.
Process:
- Belief-by-belief analysis - Shows how each of your beliefs fares across all 4 scenarios
- Specialist reactions - Domain experts comment on where your worldview might be missing something
- Crux-to-scenario mapping - Shows how your uncertainties map to scenario boundaries
- Personalized early warnings - Signals that would confirm or challenge your current view
- Reflection - Exploratory conversation about what shifted in your thinking
Output: worldview_integration.md
This is not persuasion. The goal is helping you see how different futures connect to your existing understanding.
Each specialist brings unique perspective and characteristic blind spots:
| Specialist | Focus Area | Strengths |
|---|---|---|
| Elena Vasquez | Systems Ecology | Feedback loops, interdependencies, resilience |
| Marcus Chen | Geopolitics | Power dynamics, resources, state interests |
| Aisha Okonkwo | Cultural Anthropology | Values, meaning-making, lived experience |
| Kenji Tanaka | Technology Futures | Capability thresholds, platform shifts, S-curves |
| Sarah Blackwood | Economics | Debt dynamics, financial structures, regime shifts |
| Jamie O'Sullivan | Contrarian Provocateur | Hidden risks, fragilities, challenging assumptions |
Phase 6a activates a second downstream panel. These personas do not model the world; they translate world change into consequences for a focal actor such as a household, company, customer, school system, or buying committee.
| Persona | Function |
|---|---|
| Marisol Vega | Ledger Keeper: cash flow, affordability, debt service, financial pressure |
| Darnell Brooks | Friction Mechanic: workflow drag, operational friction, execution burden |
| Nadia Rahman | Dependency Cartographer: chokepoints, institutional dependencies, transmission channels |
| Dr. Imani Clarke | Burden Cartographer: who carries cost, stress, coordination, and blame |
| Ethan Rowe | Optionality Conservator: reversibility, sequencing, lock-in, preserved options |
| Priya Desai | Precedent Archivist: structural analogs and how similar actors handled them |
| Luis Ortega | Signal Mason: decision-grade indicators, thresholds, and monitoring bricks |
| Jamie O'Sullivan | Contrarian: cross-cutting challenge role retained in the impact layer |
The impact kernel is mandatory. Overlay packs add more domain-specific seats when the question requires them.
| Overlay | Use For | Added Seats |
|---|---|---|
| household_personal | Personal finance, household decisions, partner/school exposure, college-age children, lifestyle choices | Rachel Kim, Monica Alvarez, Dr. Leah Morgan |
| commercial_positioning | Buyer-side product positioning, procurement, operations fit, implementation trust | Aarti Menon, Miguel Salazar, Hannah Stern |
shell-scenario-panel/
├── .claude/
│ ├── config.json # Project configuration
│ ├── settings.local.json # Tool permissions
│ ├── hooks/ # Quality enforcement
│ ├── scenario-init.sh # Create new scenario
│ ├── list-scenarios.sh # View all scenarios
│ ├── validate-scenario.sh # Validate scenario completeness
│ └── archive-scenario.sh # Archive completed work
├── scenarios/
│ ├── active/ # Current scenario planning
│ └── archived/ # Completed scenarios
├── prompts/
│ ├── moderator.md # Dr. Michelle Wells
│ ├── specialists/ # 6 domain specialists
│ └── impact_specialists/ # Impact translation cast
├── skills/ # Bundled Claude Code skills
│ └── worldview-elicitor/ # Phase 0 elicitation skill
├── templates/ # Scenario document templates
│ ├── worldview_model.md # Phase 0 template
│ ├── impact_analysis.md # Phase 6a template
│ ├── strategy_analysis.md # Phase 6b template
│ ├── worldview_integration.md # Phase 7 template
│ ├── focal_question.md
│ └── [other templates...]
└── CLAUDE.md # Facilitator instructions
Each scenario produces:
scenarios/active/SCENARIO-YYYY-NNN/
├── metadata.json # Tracking and status
├── worldview_model.md # Phase 0: Your mental model
├── scenario_context.md # User feedback accumulation
├── focal_question.md # Phase 1: Decision context
├── predetermined_elements.md # Phase 2: Locked-in trends
├── critical_uncertainties.md # Phase 3: Key uncertainties
├── scenarios/ # Phase 4: Future narratives
│ ├── scenario_1_[name].md
│ ├── scenario_2_[name].md
│ ├── scenario_3_[name].md
│ └── scenario_4_[name].md
├── impact_analysis.md # Phase 6a: Actor-relative translation
├── strategy_analysis.md # Phase 6b: Strategy / response testing
├── worldview_integration.md # Phase 7: Connected to your model
├── conversations/ # Specialist transcripts
│ ├── ecologist_round1_full.md
│ ├── economist_round1_full.md
│ ├── ledger_keeper_impact.md # Phase 6a
│ ├── contrarian_worldview_reaction.md # Phase 7
│ └── [etc...]
└── artifacts/ # Supporting files
The worldview-elicitor skill is required for Phase 0. It is bundled with this repository.
Installation:
# From the shell-scenario-panel directory
mkdir -p ~/.claude/skills
ln -s "$(pwd)/skills/worldview-elicitor" ~/.claude/skills/worldview-elicitorVerification:
ls ~/.claude/skills/worldview-elicitor/
# Should show: skill.mdSee skills/README.md for more details.
The system is built around understanding YOUR perspective:
- Phase 0 captures how you think before showing you anything
- Phases 1-5 explore the external world objectively
- Phase 6a translates external scenarios into actor-relative impact
- Phase 6b tests strategies or responses using that impact layer
- Phase 7 translates findings back through your lens
This means scenarios are:
- More personally relevant
- Framed in language you understand
- Connected to your existing mental models
- Explicit about where they confirm or challenge your view
The system keeps reflection explicit while forcing constructive opposition:
- Calibration: user familiarity + facilitation stance (mirror / hybrid / analytic)
- Not-knowing diagnostic: actions, outcomes, causation, value, with tool matching
- Mirror / Opposition / Synthesis: preserve input reflection and stress-test it
- Provenance logging: new inputs tracked as claims with confidence and falsifiers
- Progressive disclosure: plain-language summary first, optional technical notes second
See docs/model-guardrails-proposal.html for the full guardrail rationale and implementation map.
Moderator (Dr. Wells) enforces transcripts - hooks provide reminders only.
Multi-layer approach:
- Specialist prompts with explicit transcript requirements
- Moderator verifies transcript after each consultation
- Post-task hooks remind of responsibilities
- Metadata records all consultations with transcript paths
Automated quality analysis at key checkpoints:
- Phase 2 completion
- Phase 3 completion
- Phase 4 completion
- Phase 6a completion
- Executive Summary
Quality Analyst evaluates intellectual integrity and prevents regression to generic analysis.
# Initialize new scenario
.claude/scenario-init.sh
# List all scenarios
.claude/list-scenarios.sh
# Validate scenario completeness
.claude/validate-scenario.sh SCENARIO-2025-001
# Archive completed scenario
.claude/archive-scenario.sh SCENARIO-2025-001Scenarios are not predictions. They are:
- Plausible - Internally consistent and grounded in reality
- Challenging - Stretch thinking beyond comfortable assumptions
- Relevant - Illuminate decisions that need to be made
- Divergent - Explore genuinely different futures
- Memorable - People can recall and reason with them
- Personal - Connected to your worldview, not generic
Impact is not advice. It is the actor-relative translation layer between external scenarios and recommendations.
Worldview integration is not persuasion. It's helping you see how different futures connect to your existing understanding.
The goal is preparing minds for multiple futures, identifying actor-level consequences, stress-testing responses, and making better decisions under uncertainty.
- Be honest in Phase 0 - The more accurately we capture your worldview, the better the integration
- Embrace uncertainty - Saying "I don't know" is valuable information
- Notice resistance - Scenarios that feel implausible may reveal blind spots
- Watch your cruxes - The early warning signals for your specific uncertainties are most valuable
- Reflect genuinely - Phase 7 reflection is for you, not for us
Convert your scenario planning outputs into a presentation deck using the keynote-slides skill.
.claude/export-deck-brief.sh SCENARIO-2025-001This consolidates: focal question, predetermined elements, critical uncertainties, all four scenarios, early warning signals, impact analysis, strategy analysis, and key specialist insights.
# Create a new deck
cd ../keynote-slides-skill
scripts/new-deck.sh futures-deck --entity acme --title "Strategic Futures" --type strategy
# Copy the brief as materials
cp ../ai-simulations/shell-scenario-panel/scenarios/active/SCENARIO-2025-001/deck-brief.md \
decks/futures-deck/resources/materials/The narrative engine will help match your scenarios to storytelling frameworks like "Time Machine" (future vision) or "Rashomon" (multiple perspectives).
This system adapts:
- Shell Methodology - Classic scenario planning approach from Pierre Wack and Ted Newland
- Worldview Elicitation - Techniques from forecasting and calibration training
- Lens-World-Lens Architecture - Novel integration of internal models with external scenarios
Ready to explore multiple futures? Initialize a scenario and begin.