CritiCut is an interactive network resilience and vulnerability analysis platform that identifies the most critical nodes and fragile connections in a graph, then intelligently removes risky edges to strengthen the network without breaking it.

It blends graph theory, linear algebra, and simulation to model how real-world systems behave under failure — from communication networks and cloud infrastructure to social and transportation networks.

CritiCut takes a network and answers one powerful question:

“If this network starts to fail, where will it break first — and how can we prevent it?”

The system:

• Computes information centrality using the graph Laplacian and its pseudo-inverse to measure how important each node is to overall connectivity
• Identifies critical nodes whose instability would have the largest impact on the network
• Analyzes each connected edge using effective resistance, a metric from electrical network theory that reveals how fragile or redundant a connection is
• Removes the most vulnerable edges only if the network remains connected, reducing cascading-failure risk
• Visualizes the network before and after optimization
• Generates a detailed log of every edge removal

CritiCut does not guess.
It uses mathematical models of connectivity to understand how information, traffic, or power flows through a network — and where it is most likely to collapse.

This makes it applicable to:

• Cloud and data-center reliability
• Cybersecurity and attack-surface analysis
• Power-grid and infrastructure modeling
• Social-network influence and stability

• Laplacian-based information centrality to detect high-impact nodes
• Effective resistance to rank fragile edges
• Safe edge removal that preserves global connectivity
• Real-time graph visualization (interactive and static)
• Support for uploaded datasets and synthetic networks
• Downloadable optimization logs for full transparency

• Python – core algorithms and simulation
• NetworkX – graph modeling and traversal
• NumPy & SciPy – Laplacian matrices and linear-algebra computations
• Plotly & Matplotlib – graph visualization
• Streamlit – interactive web interface

Most graph projects simply draw networks.

CritiCut understands them.

It shows how failures propagate, which components matter most, and how to make a network more resilient — the same kind of analysis used in real-world infrastructure and reliability engineering.