The AI Agents (Your Digital Team)
Relevant source files
The following files were used as context for generating this wiki page:
The AI Network Operations Platform provides a self-hosted AI assistant that functions as a virtual network engineering team. It allows network engineers to interact with their infrastructure using plain English to perform complex tasks like discovery, troubleshooting, and configuration management across various vendors.
Sources: README.md:3-8, CLAUDE.md:3-12
Overview
The platform acts as an AI Network Engineer for enterprise infrastructure teams, solving the problem of managing multi-vendor environments through a unified conversational interface. It handles everything from simple inventory lookups to complex BGP troubleshooting and security audits.
Key Philosophies
- Local-First & Self-Hosted: All data and AI processing stay within your controlled environment to ensure security and privacy.
- Human-in-the-Loop: Every change requested by the AI requires a human engineer to approve it before execution.
- Auditability: Every decision made by an AI agent is explained and logged for full transparency.
Sources: README.md:3-8, CLAUDE.md:34-42
The Specialist Team
The platform uses "Agents" — specialized digital team members — to handle specific tasks. When you ask a question, a Master Architect (the supervisor) determines which specialist is best suited to help:
| Agent | Expertise |
|---|---|
| Discovery | Scans the network and builds the inventory of devices and neighbors. |
| Troubleshooting | Diagnoses why things are broken (e.g., routing faults, BGP drops). |
| DDI Agent | Manages DNS records and DHCP scopes/leases. |
| Security Agent | Audits firewall policies for overly-permissive or redundant rules. |
| Config Agent | Tracks configuration drift and compliance against hardening policies. |
| Documentation | Generates topology diagrams, inventories, and runbooks. |
| Packet Analysis | Analyzes pcap files to find top talkers or protocol errors. |
| Automation | Executes change requests that have already been approved by a human. |
| Consultant | Asks clarifying questions if your request is too vague. |
Sources: CLAUDE.md:44-55, backend/app/llm/prompts/init.py:316-414
Supported Vendors and Protocols
The platform supports a wide array of infrastructure through a plugin system:
- Routing/Switching: Cisco (IOS, IOS-XE, NX-OS), Juniper JunOS, Arista EOS.
- Security: Palo Alto PAN-OS, Fortinet FortiOS.
- Load Balancing & DDI: F5 BIG-IP, BlueCat, Infoblox.
- Cloud & Virtualization: AWS, Azure, VMware.
Sources: CLAUDE.md:14-27, README.md:21-25
Architecture
The platform consists of seven primary components that work together to serve the network engineer.
- Frontend: The web console where you chat with the AI, view the topology map, and approve changes.
- API: The central brain that manages sessions, handles security, and routes instructions to the agents.
- Worker: The background engine that performs the heavy lifting, such as running discovery scans, taking configuration backups, and capturing packets.
- Postgres: The system of record that stores all device data, user information, and AI reasoning logs.
- Neo4j: A living topology map that tracks how every device, interface, subnet, and DNS zone connects to everything else.
- Redis: A fast communication channel that coordinates tasks between the API and the background workers.
- Ollama: The local inference engine that runs the Large Language Models (LLMs) used by the agents.
Sources: README.md:27-38, deploy/docker/README.md:9-15
Vendor Plugin System
The platform uses standardized connectors (plugins) for each vendor. No matter if you are querying a Cisco router or an Infoblox DDI appliance, the platform normalizes the data into a consistent format. This allows the AI agents to reason about the network without needing to understand the specific CLI or API quirks of every individual device.
Sources: README.md:21-25, backend/app/agents/documentation/agent.py:12-16
Deployment Options
- Docker Compose: Ideal for lab environments or small teams. It runs the entire platform on a single server using a simple setup.
- Kubernetes/Helm: The production-grade option. It supports multi-node deployments for high availability and integrates with enterprise tools like cert-manager for automated security.
Sources: deploy/docker/README.md:3-6, deploy/kubernetes/netops/README.md:3-13
Change Approval Workflow
Safety is a core design principle. AI agents are generally read-only by default. When an agent needs to make a change, it follows a strict lifecycle:
- Draft: The agent (e.g., DDI or Security) creates a Change Request (CR) based on your request.
- Pending Approval: The CR appears in the web console for a human engineer to review.
- Pre-Change Backup: Once approved, the platform automatically takes a configuration backup of the affected devices.
- Executing: The Automation Agent pushes the changes to the infrastructure.
- Completion/Rollback: The change is marked as completed or, if it fails, the system provides the means to roll back to the previous state.
Sources: backend/app/agents/ddi/agent.py:18-24, backend/app/agents/security/agent.py:18-25, deploy/kubernetes/netops/README.md:32-35
Visual Guides
System Overview
This diagram shows how users interact with the platform and how the platform connects to your infrastructure.
Sources: CLAUDE.md:14-27, README.md:27-38
Agent Routing Flow
When you send a request, the Master Architect decides which specialist to involve.
Topology Data Model
This represents "what the platform knows" about your network in the living topology map.
Change Approval Flow
The journey of a change from the initial idea to execution on the network.
Summary
The AI Agents in this platform function as a coordinated digital team that bridges the gap between natural language requests and complex multi-vendor network operations. By utilizing a central supervisor and specialized agents, the platform ensures that tasks are routed to the appropriate expert while maintaining a strict security posture through human approval of all network changes.