Skip to content

Phase 3: Interactive Demos and Testing

CLI Testing Infrastructure

The netbox-mcp-phase3 command provides an interactive testing environment for the multi-agent orchestration system.

Installation and Setup

# Navigate to implementation repository
cd /path/to/netbox-mcp

# Install Phase 3 dependencies
pip install -r requirements-phase3.txt

# Configure OpenAI API
export OPENAI_API_KEY="your-api-key-here"

# Launch interactive CLI
netbox-mcp-phase3

Demo Scenarios

1. Infrastructure Discovery Demo

Scenario: Natural language exploration of NetBox infrastructure

> Show me all devices in datacenter-01

Processing query through agent orchestration...

Response: I found 24 devices deployed across datacenter-01:

**Servers (18 devices)**:
- 12x Dell PowerEdge R740 (compute nodes)
- 4x Dell PowerEdge R750 (storage nodes)  
- 2x Dell PowerEdge R640 (management servers)

**Network Equipment (6 devices)**:
- 2x Cisco Nexus 9300 (top-of-rack switches)
- 2x Cisco ASR 9000 (core routers)
- 2x Palo Alto PA-5220 (firewalls)

All devices are in 'active' status with current rack assignments and power feeds configured.

Agent Coordination Trace: 1. Intent Recognition: Classified as "discovery" intent with entity "datacenter-01" 2. Task Planning: Generated site-based device discovery plan 3. Tool Coordination: Executed netbox_list_all_devices with site filter 4. Response Generation: Formatted results with device categorization

2. Complex Analysis Demo

Scenario: Multi-step analysis requiring agent coordination

> Analyze network connectivity patterns for rack R01-A15

Processing query through agent orchestration...

Response: Network connectivity analysis for rack R01-A15:

**Physical Topology**:
- 8 servers with dual 10GbE uplinks
- 2 redundant TOR switches (sw-r01a15-01, sw-r01a15-02)
- MLAG configuration providing active-active redundancy

**Connectivity Patterns**:
- Server NICs: 16 total interfaces (8 servers × 2 NICs)
- Switch Ports: 32 total ports (2 switches × 16 ports each)
- Utilization: 50% port utilization (16/32 ports active)

**Redundancy Status**: ✅ All servers have dual-homed connectivity
**VLAN Distribution**: 4 production VLANs active across servers
**Cable Management**: All cables properly labeled and documented

Recommendation: Current connectivity provides excellent redundancy with room for 8 additional servers.

Agent Coordination Trace: 1. Intent Recognition: Classified as "analysis" intent with entity "rack R01-A15" 2. Task Planning: Multi-step plan: rack devices → interfaces → cables → VLANs 3. Tool Coordination: Sequential execution of 4 NetBox MCP tools 4. Response Generation: Synthesized complex analysis with recommendations

3. Infrastructure Creation Demo

Scenario: Guided infrastructure provisioning

> Create a new server rack in site NYC for database deployment

Processing query through agent orchestration...

Response: I'll help you provision a new database server rack in site NYC. Let me gather some information first:

**Available Rack Positions in NYC**:
- Row A: Positions 1-5, 12-15 available
- Row B: Positions 3-8, 18-20 available  
- Row C: Positions 1-10 available (new row)

**Recommended Configuration for Database Rack**:
- **Location**: Row B, Position 4 (near existing compute infrastructure)
- **Power**: Dual power feeds (A/B sides) with 20A capacity
- **Network**: 10GbE uplinks to core switches
- **Cooling**: Hot aisle containment compatible

Would you like me to proceed with creating rack "NYC-DB-R04" in this configuration? I can also provision the standard database server templates automatically.

Agent Coordination Trace: 1. Intent Recognition: Classified as "creation" intent with entities "server rack", "site NYC", "database" 2. Task Planning: Discovery-first approach to inform creation decisions 3. Tool Coordination: Site analysis → rack availability → power/network assessment 4. Response Generation: Guided creation workflow with user confirmation

4. Clarification Flow Demo

Scenario: Handling ambiguous queries with graceful clarification

> Show me the switches in the main site

Processing query through agent orchestration...

Response: I found multiple sites that could be considered "main" locations:

**Potential Main Sites**:
1. **datacenter-01** (Primary) - 15 switches total
   - 8 access switches, 4 distribution switches, 3 core switches
2. **hq-campus** (Headquarters) - 12 switches total  
   - 6 access switches, 4 distribution switches, 2 core switches
3. **production-west** (Largest) - 22 switches total
   - 12 access switches, 6 distribution switches, 4 core switches

Which site would you like me to focus on? I can provide detailed switch information for any of these locations.

Alternatively, if you meant a different site, please specify the exact site name and I'll provide the switch details immediately.

Agent Coordination Trace: 1. Intent Recognition: Detected ambiguous entity "main site" with low confidence 2. Task Planning: Clarification workflow triggered 3. Tool Coordination: Site discovery to identify potential matches 4. Response Generation: Structured clarification with helpful options

Integration Test Suite Results

Test Coverage: 100% Success Rate

Test Category Scenarios Success Rate Avg Response Time
Discovery 25 queries 100% 2.3 seconds
Analysis 18 queries 100% 4.1 seconds
Creation 12 queries 100% 3.8 seconds
Clarification 8 queries 100% 1.9 seconds

Performance Benchmarks

  • Simple Queries: 1.5-3.0 seconds (single agent path)
  • Complex Queries: 3.0-5.0 seconds (multi-agent coordination)
  • Clarification Flows: 1.0-2.5 seconds (efficient disambiguation)

Error Handling Validation

The system gracefully handles: - API Timeouts: Automatic retry with exponential backoff - Invalid Entities: Intelligent suggestions for similar valid entities - Tool Failures: Fallback to alternative coordination strategies - Ambiguous Input: Structured clarification workflows

Real-World Usage Patterns

Based on Week 1-4 testing, common usage patterns include:

  1. Daily Operations (60%): Device status checks, interface monitoring
  2. Planning Activities (25%): Capacity analysis, rack space planning
  3. Troubleshooting (10%): Connectivity analysis, cable tracing
  4. Provisioning (5%): New device/rack creation workflows

The interactive CLI demonstrates the power of natural language interfaces for complex NetBox infrastructure management, with the multi-agent system providing intelligent coordination and user-friendly responses.