Quick Prompts

Overview

Quick Prompts are pre-built AI prompt templates that standardize common network operations questions. Instead of typing the same diagnostic question every time a BGP neighbor goes down or an interface starts dropping packets, you create a reusable template with variable placeholders that get filled in at execution time.

Quick Prompts support variable placeholders using the {{variable_name}} syntax. When you execute a prompt, NetStacks asks you to fill in each variable (or auto-populates it from context like the currently selected device) before sending the completed prompt to AI Chat.

• Reusable templates — Write a prompt once, use it across any device or interface
• Variable placeholders — Dynamic values like device names, IP addresses, and interface IDs filled in at execution time
• Org-level sharing — Share approved prompts with your entire team so everyone uses consistent troubleshooting workflows
• Favorites — Mark frequently used prompts as favorites for quick access

How It Works

The Quick Prompts system is built around a template storage and execution model:

Template Storage

Quick Prompts are stored in PostgreSQL, scoped to an organization. Each prompt has a name, the template body with variable placeholders, a favorite flag, and a shared flag that controls visibility. Prompts are owned by the user who created them.

Variable Placeholder Syntax

Variables use double curly braces: {{variable_name}}. Variable names should be descriptive and use underscores for multi-word names (e.g., {{device_name}}, {{neighbor_ip}}, {{interface}}). NetStacks extracts all unique variables from the template body and presents them as input fields when the prompt is executed.

Variable Resolution

When you execute a Quick Prompt, variables can be filled in two ways:

• Manual input — You type the value into an input field for each variable
• Auto-populated from context — If you have a device selected in the Terminal, variables like {{device_name}} and {{device_ip}} are automatically filled in

Visibility Scoping

Each prompt has a shared flag. When set to false, only the creator can see and use the prompt. When set to true, all users in the organization can see and execute it. Only the creator can edit or delete a shared prompt.

Execution Flow

1. Select a Quick Prompt from your library or the shared library
2. Fill in any variable placeholders (some may auto-populate)
3. Preview the completed prompt
4. Send to AI Chat for processing
5. AI responds with analysis, commands, or recommendations

Creating Quick Prompts

Follow these steps to create a new Quick Prompt template.

Step 1: Navigate to Quick Prompts

Go to AI → Quick Prompts in the Controller web interface or access the Quick Prompts panel from the AI Chat sidebar.

Step 2: Create a New Prompt

Click Create Prompt and enter a descriptive name that clearly identifies the use case (e.g., "Troubleshoot BGP Neighbor" or "VLAN Audit Report").

Step 3: Write the Prompt Template

Write the prompt body using natural language and include variable placeholders where dynamic values should be inserted. Be specific in your instructions to get the best AI responses.

Step 4: Set Favorite and Sharing

Mark the prompt as a favorite if you use it frequently — favorites appear at the top of your prompt list. Toggle sharedto make the prompt visible to all users in your organization.

Step 5: Test the Prompt

Execute the prompt with sample variable values to verify the AI produces useful results. Iterate on the wording until you are satisfied with the output quality.

Step 6: Save

Click Save. The prompt is immediately available in your library (and the shared library if sharing is enabled).

Code Examples

BGP Neighbor Troubleshooting Prompt

A comprehensive diagnostic prompt for investigating BGP neighbor issues:

{
  "name": "Troubleshoot BGP Neighbor",
  "prompt": "Analyze the BGP neighbor {{neighbor_ip}} on device {{device_name}}. Check the current state, uptime, and prefixes received. If the neighbor is not in Established state, identify the likely cause (authentication mismatch, AS number mismatch, network reachability, hold timer expiry) and suggest specific commands to diagnose and fix the issue. Include the exact CLI commands for {{platform}} (e.g., Cisco IOS-XE, Juniper Junos).",
  "is_favorite": true,
  "shared": true
}

VLAN Audit Prompt

Compare device VLANs against your organization's standard plan:

{
  "name": "VLAN Audit",
  "prompt": "List all VLANs configured on {{device_name}} and compare them against our standard VLAN plan. Our standard VLANs are: 10 (Management), 20 (Users), 30 (VoIP), 40 (Printers), 100 (Servers), 999 (Native/Unused). Identify any non-standard VLANs, missing required VLANs, and VLANs assigned to trunk ports that should not carry them.",
  "is_favorite": false,
  "shared": true
}

Interface Health Check Prompt

Detailed interface diagnostics for error investigation:

{
  "name": "Interface Health Check",
  "prompt": "Check interface {{interface}} on {{device_name}} for the following: input/output error counters, CRC errors, frame errors, input/output packet rates, bandwidth utilization percentage, duplex and speed settings, and any recent interface flaps. Provide a health assessment (Healthy, Warning, Critical) with specific recommendations if issues are found.",
  "is_favorite": true,
  "shared": true
}

Variable Placeholder Examples

Common variables used across Quick Prompts:

# Device context variables (can auto-populate)
{{device_name}}      # e.g., "core-rtr-01.dc1"
{{device_ip}}        # e.g., "10.0.1.1"
{{platform}}         # e.g., "Cisco IOS-XE"

# Interface variables
{{interface}}        # e.g., "GigabitEthernet0/0/1"
{{interface_ip}}     # e.g., "192.168.1.1/30"

# Protocol variables
{{neighbor_ip}}      # e.g., "10.255.0.2"
{{as_number}}        # e.g., "65001"
{{vlan_id}}          # e.g., "100"

# General variables
{{site_name}}        # e.g., "DC1-East"
{{ticket_number}}    # e.g., "INC0012345"

Creating a Prompt via API

curl -X POST https://controller.example.net/api/quick-prompts \
  -H "Authorization: Bearer <token>" \
  -H "Content-Type: application/json" \
  -d '{
    "name": "OSPF Neighbor Check",
    "prompt": "Show all OSPF neighbors on {{device_name}} and verify they are in FULL state. For any neighbors not in FULL state, explain the current state and what it means. Suggest diagnostic commands for {{platform}}.",
    "is_favorite": false,
    "shared": true
  }'

Questions & Answers

Q: What are Quick Prompts?

A: Quick Prompts are reusable AI prompt templates stored in NetStacks. Each template contains a natural language prompt with variable placeholders (like {{device_name}}) that get filled in when you execute the prompt. They standardize common network operations questions so your team asks consistent, thorough diagnostic questions every time.

Q: How do variable placeholders work?

A: Variables use the {{variable_name}} syntax inside the prompt body. When you execute a Quick Prompt, NetStacks extracts all unique variables and presents them as input fields. You fill in the values (or they auto-populate from context), and the completed prompt is sent to AI Chat. For example, {{device_name}} might be replaced with "core-rtr-01.dc1" at execution time.

Q: Can I share prompts with my team?

A: Yes. Each Quick Prompt has a shared flag. When enabled, the prompt is visible to all users in your organization. Only the creator can edit or delete shared prompts. This allows senior engineers to build a library of approved troubleshooting workflows that the entire team can use.

Q: How do I use a Quick Prompt in AI Chat?

A: Open the Quick Prompts panel from the AI Chat sidebar, select a prompt from your personal or shared library, fill in the variable values, preview the completed prompt, and click Send. The prompt is submitted to AI Chat as if you had typed it manually, and the AI responds with its analysis.

Q: Can variables be auto-populated from device context?

A: Yes. If you have a device selected in the Terminal session when you open a Quick Prompt, variables like {{device_name}}, {{device_ip}}, and {{platform}} are automatically filled in based on the current device context. You can override auto-populated values before executing the prompt.

Q: What makes a good Quick Prompt template?

A: Good templates are specific about what to analyze, include context about the expected environment, and request a structured output format. Instead of "Check BGP on this router," write a prompt that specifies the neighbor IP, asks for specific metrics (state, uptime, prefixes), and requests platform-specific CLI commands. The more detail you provide, the more useful the AI response will be.

Troubleshooting

Variables not resolving

If variable placeholders appear in the final prompt as literal text (e.g., you see {{device_name}} instead of the actual device name):

• Verify the placeholder syntax uses double curly braces: {{variable_name}}
• Check that you filled in all required variable fields before executing
• For auto-populated variables, ensure a device is selected in the active Terminal session

Prompt producing poor results

If the AI response is not useful:

• Add more context to the prompt template — specify the platform, the exact metrics you want analyzed, and the output format you expect
• Break complex prompts into multiple focused prompts rather than one large one
• Include example output or reference documentation in the prompt body

Shared prompts not visible to team members

If a shared prompt does not appear for other users:

• Verify the shared flag is set to true on the prompt
• Confirm the other users are in the same organization
• Check that the users have permission to access AI features in their role configuration

Related Features

Quick Prompts work alongside other NetStacks AI capabilities:

• AI Chat — The interface where Quick Prompts are executed and responses are displayed
• Quick Actions — One-click API call templates that execute real operations (compared to Quick Prompts which send text to AI)
• NOC Agents — Autonomous agents that can use prompt patterns for systematic diagnostics
• Command Suggestions — Real-time command suggestions that complement template-based prompts
• Knowledge Base — Documentation and runbooks that AI references when answering prompts