Automated Documentation
Relevant source files
The following files were used as context for generating this wiki page:
Automated Documentation is a core capability of the AI Network Operations Platform that provides network engineers with instantly generated, up-to-date technical artifacts. Instead of manually updating spreadsheets or drawing Visio diagrams, the platform uses its "living topology map" and live discovery data to produce inventories, topology diagrams, and operational runbooks on demand.
The system is designed to be "local-first" and secure, ensuring that sensitive network information remains within the organization's infrastructure. Every piece of documentation is grounded in live data, meaning it reflects the actual state of the network rather than a stale design document. For security, any sensitive data like community strings or passwords are automatically redacted (hidden) before being processed by AI models.
Sources: CLAUDE.md:1-20, README.md:1-10, backend/app/agents/documentation/agent.py:1-20
Overview
The platform serves as an AI Network Engineer that helps operations teams manage complex, multi-vendor environments. It solves the problem of documentation debt by automating the creation of critical artifacts that usually require hours of manual labor to maintain.
Key Capabilities and Principles
- Multi-Vendor Support: Works across Cisco (IOS, XE, NX-OS), Juniper, Arista, Palo Alto, Fortinet, F5, and DDI providers like Infoblox and BlueCat.
- Specialized AI Agents: The platform uses "specialized team members" (Agents). When you ask for a diagram, the Master Architect (the supervisor) routes your request to the Documentation Agent.
- Local and Secure: The platform is self-hosted. All AI reasoning is visible and audited, and every configuration change requires human approval.
- Read-Only Operations: The documentation system is strictly read-only; it observes the network but never changes device settings.
Sources: CLAUDE.md:14-38, README.md:1-15, backend/app/agents/documentation/agent.py:45-65
System Connectivity
This diagram shows how the platform interacts with your existing network infrastructure and external providers.
Architecture
The platform consists of seven functional components that work together to discover the network and generate documentation.
Platform Components
- frontend: The operations console where you chat with the AI, view the topology map, and manage approvals.
- api: The central brain that handles requests, checks permissions, and manages agent sessions.
- worker: The background engine that runs discovery jobs, builds documentation, and captures packets without blocking your session.
- postgres: The system of record that stores device inventories and historical data.
- neo4j: A living topology map that tracks how every device, interface, subnet, and DNS zone connects to everything else.
- redis: A high-speed messenger used to manage background tasks and temporary data.
- ollama: An optional local AI engine that allows the platform to process data without sending anything to the cloud.
Sources: README.md:20-30, deploy/docker/README.md:5-15
The Vendor Connector System
To handle different manufacturers, the platform uses a standardized connector (Plugin System) for each vendor. Whether the platform is talking to a Cisco router or a BlueCat Address Manager, the connector normalizes the output into a consistent format before the Documentation Agent uses it.
Sources: README.md:32-40, backend/app/agents/documentation/agent.py:15-25
Deployment Options
Network teams can choose between two primary deployment methods depending on their scale.
| Feature | Docker Compose | Kubernetes (Helm) |
|---|---|---|
| Use Case | Labs, small teams, or single-server setups. | Production environments and high-availability needs. |
| Complexity | Simple: single command to start. | Higher: requires a cluster and more configuration. |
| Scale | Vertical (one large server). | Horizontal (multi-node cluster). |
| Availability | Single points of failure. | Resilient: services can restart on different nodes. |
Sources: deploy/docker/README.md:1-10, deploy/kubernetes/netops/README.md:1-15
Documentation Artifacts
The Documentation Agent acts as a specialized team member focused on creating artifacts. It strictly follows a "Read-Only" rule, ensuring it cannot accidentally change network configurations.
1. Network Inventories
Inventories are generated deterministically (exactly as the data exists) in Markdown or CSV formats. They include:
- Devices: Hostnames, Management IPs, Vendors, and Serial Numbers.
- Interfaces: Names, Descriptions, IP addresses, and Status.
- Neighbors: Connectivity data derived from LLDP or CDP.
- Routes: Routing table entries including prefixes and next-hops.
Sources: backend/app/agents/documentation/agent.py:10-15, backend/tests/agents/documentation/test_documentation_agent.py:40-100
2. Topology Diagrams
The platform generates Mermaid-format diagrams directly from the Neo4j living topology map. These diagrams show the relationships between various network entities.
What the platform knows about your network:
3. Operational Runbooks
Runbooks combine raw network facts with an AI-generated narrative. The AI explains "Operational Procedures" and "Overview" while the platform injects data tables for accuracy. Before the AI sees any data, the platform redacts sensitive information.
Sources: backend/app/agents/documentation/agent.py:19-24, backend/tests/agents/documentation/test_documentation_agent.py:650-680
Process and Routing
When an engineer interacts with the platform, the request follows a specific path to ensure the right agent handles it and all changes are authorized.
Agent Routing Workflow
The "Master Architect" decides which specialist to involve based on your goal.
Change Approval Lifecycle
While documentation is read-only, other agents might propose changes. Documentation often serves as the "Pre-change" or "Post-change" evidence in this flow.
Sources: README.md:10-15, deploy/docker/README.md:55-65
Conclusion
Automated Documentation transforms the Neo4j "living topology map" into actionable business intelligence. By routing requests through the specialized Documentation Agent, the platform ensures that network inventories, Mermaid diagrams, and operational runbooks are always grounded in live, normalized data, while maintaining a strict security posture through automated secret redaction.
Sources: backend/app/agents/documentation/agent.py:1-25, backend/tests/agents/documentation/test_documentation_agent.py:1-20