Negative Space Framework

An epistemology of absence for AI reasoning. The Negative Space Framework shifts AI focus from interpolation to "Void Mapping"—identifying and classifying gaps (informational, causal, dependency) between current states and desired goals.

Traditional engineering focuses on assembly. Negative Space Reasoning maps what's missing—transforming systemic uncertainty into precisely mapped territory for agentic navigation.

Read the full manifesto →

🧭 The Logic Flow (The Void Engine)

graph TD
    A[Point A: Current Reality] --> V{Void Mapping Engine}
    B[Point B: Target Objective] --> V
    V --> G[Gap Characterization]
    G --> C[Dimensional Cluster Analysis]
    C --> N[Strategic Navigation Plan]
    N --> D[Directed Capability Acquisition]
    D --> A

Loading

Figure 1: The Void Engine identifies dependency gaps, information voids, and constraint barriers through contrastive analysis rather than interpolation.

🚀 Installation

From PyPI (Recommended)

pip install negative-space

From Source

# Clone the repository
git clone https://github.com/WADELABS/negative-space.git
cd negative-space

# Install in development mode
pip install -e .

# Install with development dependencies
pip install -e ".[dev]"

Requirements:

• Python 3.8 or higher
• numpy, networkx, matplotlib, dataclasses-json

🛰️ The "Now": Mapping Architectural Voids

The following snippet demonstrates use of the VoidAgent to perform a rigorous assessment of the gap between a local environment and a production-ready Kubernetes deployment.

Note: The examples below assume the package has been installed using pip install -e . or pip install negative-space.

from negative_space import VoidAgent, GapCriticality

# Initialize the observer
agent = VoidAgent(name="EpistemicMapper", rigor=0.95)

# Define the reality gap
reality_a = {"infra": "local", "security": "basic"}
objective_b = {"infra": "k8s_prod", "security": "zero_trust"}

# Generate the Void Report
report = await agent.map_voids(reality_a, objective_b)

critical_voids = [g for g in report['critical_findings'] if g['criticality'] == "BLOCKING"]
print(f"Mapped {len(critical_voids)} blocking voids in the production path.")

🎯 Running Examples

After installation, you can run the example demos:

# Featured demo (recommended) - Comprehensive microservices migration scenario
python examples/featured_demo.py

# Basic demo
python examples/demo.py

# Advanced portfolio demo
python examples/portfolio_demo.py

# Quick start
python examples/quickstart.py

See examples/README.md for detailed documentation of all examples.

📊 Strategic Metrics

Metric	Purpose	Agent Insight
Void Density	Volume of missing logic	"The path to B requires 80% new logic acquisition."
Gap Criticality	Impact of the void	"This information gap blocks all downstream causal links."
Navigability	Ease of traversing gaps	"The void is highly connected; addressing Gap X resolves Y."
Fillability	Feasibility of closure	"This constraint is emergent and may require B-redefinition."

🔧 Implementation Status

Feature	Status	Description
Core Framework
VoidAgent & VoidCollective	✅ Implemented	Agentic void mapping
Gap Classification (8 types)	✅ Implemented	VoidType enumeration
Gap Criticality Analysis	✅ Implemented	5-level criticality system
Discovery Methods
Contrastive Analysis	✅ Implemented	Compare A vs B states
Dependency Walk	✅ Implemented	Traverse dependency chains
Constraint Propagation	✅ Implemented	Identify constraint violations
Counterfactual Exploration	✅ Implemented	"What-if" gap discovery
Boundary Probing	✅ Implemented	Edge case detection
Metrics
Void Density	✅ Implemented	Weighted gap volume (0-1)
Gap Criticality	✅ Implemented	BLOCKING/HIGH/MEDIUM/LOW
Navigability	✅ Implemented	Path traversability metric
Fillability	✅ Implemented	Gap closure feasibility
Connectivity	✅ Implemented	Inter-gap network density
Navigation
Gap Hopping Strategy	✅ Implemented	Sequential gap filling
Boundary Skirting	✅ Implemented	Avoid blockers
Void Bridging	✅ Implemented	Direct path finding
Constraint Circumvention	✅ Implemented	Work around constraints
Analysis
Gap Clustering	✅ Implemented	Semantic/structural/strategic
Void Visualization	✅ Implemented	Network & distribution plots
Pattern Recognition	✅ Implemented	Historical void analysis
Planned Features
Real-time Void Tracking	🔄 Planned	Live gap monitoring
Multi-agent Consensus	🔄 Planned	Enhanced collective mapping
Automated Test Generation	🔄 Planned	Gap-driven test synthesis
Integration APIs	🔄 Planned	CI/CD, Jira, GitHub

Legend: ✅ Implemented | 🔄 Planned | ⚠️ Experimental

---

docs: formalize high-fidelity documentation and epistemological grounding Developed for WADELABS AI Safety Research 2026