Deployment AI Advisor

AI-powered assistant system for Salt Security customers to deploy, troubleshoot, and manage traffic collection across cloud environments through Claude Code CLI interactions.

Quick Start

Prerequisites

• Claude Code CLI installed and configured

Usage

Method 1: Natural Language Orchestrator

Start Claude in the project directory to automatically discover all subagents:

# Navigate to project directory and start Claude
cd deployment_ai_advisor
claude

All subagents are automatically available! Ask questions in natural language:

• "What collector should I use for AWS API Gateway?"
• "I'm getting 403 errors after deployment, how do I fix this?"
• "Validate my current setup against best practices"
• "Generate a deployment SOW for Azure APIM"
• "Use the deployment-advisor subagent to recommend collectors"
• "Have the error-handler subagent investigate this 403 error"

Method 2: Slash Commands (Built-in CLI Features)

The project includes proper Claude Code slash commands with autocompletion:

# Navigate to project directory and start Claude
cd deployment_ai_advisor
claude

Available slash commands:

• /advisor:advise "AWS API Gateway setup" --detail-level comprehensive
• /advisor:troubleshoot "403 permission errors" --cloud-provider aws --verbose
• /advisor:validate "deployment completeness" --validation-depth comprehensive
• /advisor:report "sow" --format markdown --include-diagrams

Command Discovery:

• Type / in Claude Code CLI to see all available commands with autocompletion
• Commands are stored in .claude/commands/advisor/ directory
• Each command has built-in argument hints and descriptions

Method 3: Subagent Management

Use the built-in /agents command to manage subagents:

# Navigate to project and start Claude
cd deployment_ai_advisor
claude

# View and manage all available subagents
> /agents

Available Subagents:

• orchestrator - Central controller for multi-agent coordination
• deployment-advisor - SME for collector deployment planning
• data-extractor - Centralized data retrieval with Document360 access
• error-handler - Troubleshooting specialist for error pattern matching
• validator - Deployment verification and SOW comparison
• reporter - SOW generation and session management

What You Get

Deployment Recommendations:

• Specific collector types and configurations
• Step-by-step implementation procedures
• Prerequisites and resource requirements
• Architecture diagrams and visual guides

Troubleshooting Support:

• Error pattern matching and classification
• Diagnostic commands and procedures
• Solution recommendations with success probabilities
• Escalation guidance when needed

Validation Reports:

• Deployment completeness assessment
• Gap analysis with remediation plans
• Security and compliance checking
• Performance validation

Professional Documentation:

• Statement of Work (SOW) generation
• Architecture diagrams with Mermaid
• Implementation timelines
• Risk assessments and mitigation strategies

Architecture