CyberCortex

Inspiration We were frustrated by the massive boilerplate and complexity of traditional microservices in multi-agent systems. We wanted to build an autonomous cybersecurity platform where security agents natively "live" within the network's architecture. Jaclang's Data Spatial Programming inspired us to model networks not as static tables, but as living graphs that agents can physically traverse.

What it does CyberCortex is an autonomous security validation platform. Instead of linear scripts, it models your entire network and threat landscape as an interconnected graph. Autonomous "Walkers" (Jaclang agents) seamlessly traverse this spatial data graph, scanning nodes for vulnerabilities, cross-referencing active threat intelligence, and reporting real-time security postures to a centralized dashboard.

How we built it We built the platform entirely from scratch using Jaclang.

Graph & Nodes: We designed graph.jac to define SecurityTarget and ThreatIntel nodes connected by spatial edges. Walkers: In agents.jac, we created ScannerWalker and CoordinatorWalker that natively walk the graph to execute security sweeps. Backend/UI: We used Jaclang’s seamless Python interoperability in main.jac to spin up a FastAPI server that exposes our execution environment and serves a sleek, real-time HTML/JS dashboard. Challenges we ran into The biggest challenge was shifting our mental model. Moving away from traditional REST API architectures and Object-Oriented Programming to Jaclang’s Data Spatial Programming required a completely new way of thinking. Designing the graph edges and teaching the walkers how to correctly traverse and carry state across nodes took intense iteration.

Accomplishments that we're proud of We are incredibly proud of building a 100% native Jaclang architecture. We successfully eliminated the need for heavy frameworks like Next.js or complex Docker orchestrations. Seeing our first ScannerWalker successfully traverse the spatial graph and stream vulnerabilities to the dashboard in real-time was a massive win for the team.

What we learned We learned the sheer power of Data Spatial Programming. By maintaining the state inherently within the graph structure rather than in external databases, multi-agent coordination becomes frictionless. Jaclang significantly reduces boilerplate and makes AI agent orchestration feel natural.

What's next for CyberCortex Our next step is integrating LLMs directly into the Jaclang walkers using Jaclang's native AI skills, allowing the walkers to not just find vulnerabilities, but autonomously patch them. We also plan to expand the graph to ingest and map out entire AWS/GCP cloud environments automatically.

Built With

• css
• fastapi
• html
• jaclang
• javascript
• python
• tailwind