About This Project

The Intelligent Networking Layer

This project bridges the gap between natural language and structured network infrastructure management. By leveraging Llama 3.1 via Cloudflare Workers AI, the application acts as an autonomous "ChatOps" agent for NetBox. Instead of manually navigating complex web forms or writing curl commands, users can simply state their intent—such as "provision the next available IP for a web server"—and the agent handles the underlying logic, validation, and API execution in real-time.

The Agentic Workflow: Reasoning & Execution

The core of the system is a multi-stage reasoning loop. When a user submits a prompt, the system first performs Intent Detection, where the LLM determines which specific NetBox tool (querying, provisioning, or updating) is required. Once the tool is selected, the middleware executes the request against the NetBox REST API using a secure token-based architecture. This ensures that the AI never has direct access to the database; it must work through the same business logic and permissions as a human administrator.

Data Synthesis and Feedback

Once the network action is complete, the raw JSON response from NetBox is fed back into the model for Synthesis. This is a critical step that transforms technical data into human-readable confirmation. The "Architect" reviews the operation's success or failure and explains the outcome—confirming specific IP assignments or explaining prefix exhaustion errors. This creates a closed-loop system where the network state is not just modified, but clearly communicated back to the user.

Technical Highlights

• Edge Native: Built on the Next.js App Router and deployed to Cloudflare Pages, ensuring minimal latency.
• Function Calling: Implemented sophisticated tool-calling logic to map conversational context to specific API endpoints.
• Stateful UI: A custom React frontend that manages "In-flight" snapshots of user prompts to provide a seamless, non-blocking interface during complex network operations.