grazianoguiducci/scenario-projector

Scenario Projector — Structural Decision Support

Not a prediction tool. A structural lens. The structure contains the answer — don't interpose numbers between structure and decision.

When to use

• Before a strategic decision with competing tensions
• When cross-checking a plan, thesis, or roadmap
• When prioritizing with limited resources
• When mapping risks and their correlations
• When the CEC or Domandatore reveals 5+ tensions and you need to see the field

How it works

Every tension is a dipole D(claim, anti-claim). The anti-claim isn't negation — it's the other pole. Dipoles resonate (binary: yes/no). The trajectory follows maximum potential.

4 structural lenses:

| Lens | What it reveals | |------|-----------------| | Focus | Where tensions converge — natural action points | | Leverage | Structural pillars — invest here, it propagates | | Risk | Anti-claim contradicts a neighbor — crack in the thesis | | Blind spot | Isolated from the field — unexplored or noise |

Usage

From Python

import sys
# Find scenario_projector.py in the seed scripts directory
# Typical location: d-nd-seed/plugins/d-nd-core/scripts/
sys.path.insert(0, '/path/to/d-nd-seed/plugins/d-nd-core/scripts')
from scenario_projector import ScenarioProjector

# From tensions directly
sp = ScenarioProjector(seed_data={
    'context': 'startup_strategy',  # enables domain-specific language
    'direction': 'What we are trying to achieve',
    'tensions': [
        {'id': 'TENSION_ID', 'claim': 'The thesis but the counter-thesis'},
        # ... 5-15 tensions
    ]
})

# The 4 lenses
checks = sp.cross_check()        # per-tension structural verdict
strat = sp.strategy()             # focus, leverage, risks, blind spots
plan = sp.action_plan()           # prioritized actions with domain language
result = sp.explore(verbose=True) # full trajectory + passages + field

From the command line

# Cross-check the field
python scenario_projector.py --seed path/to/seed.json --cross-check

# Strategy insights
python scenario_projector.py --seed path/to/seed.json --strategy

# Action plan
python scenario_projector.py --seed path/to/seed.json --action-plan

# Full exploration
python scenario_projector.py --seed path/to/seed.json --explore

Pre-configured domains

Seeds in d-nd-seed/plugins/d-nd-core/scripts/examples/:

• startup_strategy.json — Series A founder decisions
• product_roadmap.json — Feature prioritization
• due_diligence.json — Investment thesis cross-check
• risk_assessment.json — Digital transformation risks
• portfolio_management.json — Multi-asset rebalancing

Automated integration

See automation_pattern.py in examples/ for the full pattern:

DATA SOURCE → TensionGenerator → ScenarioProjector → ActionConsumer → EXECUTION

The projector is the cognitive middleware: domain-agnostic structural analysis between domain-specific input (tension generation) and output (action execution).

Relationship with other tools

| Tool | Role | When | |------|------|------| | CEC | How to think about each tension | During — deepens individual analysis | | Domandatore | Generates tensions from a question | Before the projector — creates the input | | Godel | Inverts a specific claim (det=-1) | On contested tensions — flip the assumption | | Projector | Maps the field of all tensions | After — sees the structural whole |

Natural flow: Domandatore → Projector → CEC on focus areas → Godel on contradictions.

Writing good tensions

• Include both poles with "but", "however", "while", "yet", or "—"
• Use concrete numbers when available
• Keep IDs semantic — PRICING_MODEL not T1
• 8-15 tensions is the sweet spot

Verdicts

| Verdict | Meaning | Action | |---------|---------|--------| | confirmed | Structural pillar with high support | Amplify — investment propagates | | supported | Backed by neighbors, not a pillar | Monitor — it's healthy | | contested | Anti-claim matches a neighbor's claim | Investigate — there's a crack | | weak | Few connections, not isolated | Strengthen or deprioritize | | unverifiable | Isolated — no neighbors to verify | Explore — could be hidden potential |

Eval

Trigger Tests

"analyze this strategy structurally" -> activates

"I have 8 competing priorities, which ones matter?" -> activates

"cross-check this thesis" -> activates

"map the risks in this plan" -> activates

"what's the git status" -> does NOT activate

"fix this bug" -> does NOT activate

Fidelity Tests

Given 5+ tensions: produces dipole field, trajectory, structural verdicts

Given a domain seed: uses domain-specific labels in action plan

Given a single tension: cross-checks it against the field

Given a mechanical task: does NOT activate, suggests direct execution