Snow-Flow CLI Commands

Snow-Flow provides a powerful CLI for ServiceNow development orchestration.

Core Commands

Starting Snow-Flow

bash

# Start interactive TUI
snow-flow

# Start with specific model
snow-flow --model claude-sonnet
snow-flow --model claude-opus
snow-flow --model gpt-4

# Resume previous session
snow-flow --resume
snow-flow --continue

System Status

bash

# Check system status
snow-flow status

# Real-time monitoring
snow-flow monitor

# Version info
snow-flow --version

SPARC Modes

SPARC (Specification, Pseudocode, Architecture, Refinement, Completion) modes provide structured development workflows.

Available Modes

Mode	Command	Purpose
Orchestrator	`sparc`	Multi-agent coordination
Coder	`sparc run coder`	Direct implementation
Researcher	`sparc run researcher`	Investigation and analysis
TDD	`sparc tdd`	Test-driven development

Usage

bash

# Orchestrator mode (default)
snow-flow sparc "Create incident dashboard widget"

# Specific mode
snow-flow sparc run coder "Implement auto-assignment business rule"
snow-flow sparc run researcher "Analyze current incident workflow"

# Test-driven development
snow-flow sparc tdd "Add SLA breach notification"

Agent Management

Spawning Agents

bash

# Spawn specialized agent
snow-flow agent spawn developer
snow-flow agent spawn researcher
snow-flow agent spawn reviewer

# List active agents
snow-flow agent list

# Agent status
snow-flow agent status <agent-id>

Agent Types

Type	Purpose
`developer`	ServiceNow artifact creation
`researcher`	Investigation and analysis
`reviewer`	Code review and validation
`tester`	Testing and QA

Swarm Coordination

Multi-agent swarms for complex tasks.

bash

# Start swarm with objective
snow-flow swarm "Build HR portal with self-service features"

# Swarm options
snow-flow swarm "objective" --strategy parallel
snow-flow swarm "objective" --strategy sequential
snow-flow swarm "objective" --mode development
snow-flow swarm "objective" --monitor

Swarm Strategies

Strategy	Description
`parallel`	Agents work simultaneously
`sequential`	Agents work in order
`adaptive`	Dynamic coordination

Task Management

bash

# Create task
snow-flow task create "Implement feature X"

# List tasks
snow-flow task list

# Task status
snow-flow task status <task-id>

# Orchestrate complex task
snow-flow task orchestrate "Multi-step implementation"

Memory Operations

Persistent memory across sessions.

bash

# Store data
snow-flow memory store <key> <data>

# Retrieve data
snow-flow memory get <key>

# List all keys
snow-flow memory list

# Search memory
snow-flow memory search <query>

# Clear memory
snow-flow memory clear

Configuration

bash

# Configure ServiceNow instance
snow-flow config instance <url>

# Set credentials
snow-flow config auth

# View configuration
snow-flow config show

# Reset configuration
snow-flow config reset

Authentication

bash

# Authenticate with ServiceNow
snow-flow auth servicenow

# Authenticate with enterprise services
snow-flow auth enterprise

# Check auth status
snow-flow auth status

# Logout
snow-flow auth logout

Environment Variables

Variable	Purpose
`SNOWCODE_MODEL`	Default AI model
`SNOWCODE_DEBUG_TOKENS`	Enable token debugging
`SNOWCODE_LOG_LEVEL`	Logging verbosity
`SERVICENOW_INSTANCE`	ServiceNow instance URL

Common Workflows

Starting Development Session

bash

# 1. Start snow-flow
snow-flow

# 2. In TUI, create Update Set first
# 3. Develop features
# 4. Complete Update Set when done

Debugging Token Usage

bash

# Enable token debugging
SNOWCODE_DEBUG_TOKENS=true snow-flow

# Check debug output
cat .snow-code/token-debug/debug-*.jsonl | jq .

Multi-Agent Development

bash

# Start swarm for complex feature
snow-flow swarm "Build customer portal" --strategy parallel --monitor

# Monitor progress
snow-flow monitor

TUI Shortcuts

Inside the interactive TUI:

Key	Action
`Ctrl+C`	Cancel current operation
`Ctrl+D`	Exit
`Tab`	Autocomplete
`Up/Down`	Navigate history
`/help`	Show help
`/clear`	Clear screen
`/debug`	Toggle debug mode

Best Practices

•Always start with Update Set - Track all changes
•Use SPARC modes - Structured development
•Monitor tokens - Watch context usage
•Persist memory - Save important data
•Use swarms - For complex multi-part tasks