Skip to main content

​
Overview

NikCLI’s agent management system provides control over two registered AI agents, enabling autonomous development with intelligent task orchestration. The system includes the Universal Agent for comprehensive full-stack development and the VM Agent for isolated, secure development environments.

Universal Agent

All-in-one enterprise agent with 44 comprehensive full-stack capabilities

VM Agent

Secure virtualized development with Docker isolation and 20 capabilities

Custom Blueprints

Create and save agent configuration blueprints

Cognitive Orchestration

Intelligent task understanding and routing
NikCLI uses two registered agents rather than multiple specialized agents. Both agents handle all development domains through intelligent cognitive routing and tool orchestration.

​
Core Agent Commands

​
/agents - List Available Agents

Display all available agents and their current status. 
# List all available agents
/agents
Output includes:
  • Agent name and specialization
  • Current status (active/idle)
  • Basic capabilities overview
Available Agents:
  • universal-agent - Full-stack development with 44 capabilities (default)
  • vm-agent - Secure virtualized development with Docker isolation (20 capabilities)

​
/agent - Execute with Specific Agent

Execute tasks using a specific agent.
  • Universal Agent
  • VM Agent
# Execute with universal agent
/agent universal-agent "create a React login component"
/agent universal-agent "optimize database queries"
/agent universal-agent "implement API authentication"
Universal Agent - 44 Capabilities:
  • Core (7): code-generation, code-analysis, code-review, optimization, debugging, refactoring, testing
  • Frontend (11): React, Next.js, TypeScript, JavaScript, HTML, CSS, frontend, components, hooks, JSX, TSX
  • Backend (8): Node.js, API development, database, server architecture, REST API, GraphQL, microservices
  • DevOps (8): CI/CD, Docker, Kubernetes, deployment, infrastructure, monitoring, security
  • Autonomous (5): File operations, project creation, autonomous coding, system admin, full-stack dev
  • Analysis (5): Performance, security, quality assessment, architecture review, documentation

​
/auto - Autonomous Execution

Enable autonomous mode for complex tasks with minimal supervision. 
# Simple autonomous task
/auto "create a todo application with React and Node.js"

# Complex autonomous project
/auto "build an e-commerce platform with authentication and payment processing"

# Autonomous optimization
/auto "identify and fix all performance issues in the project"
Autonomous Features:
  • Automatic task breakdown
  • Progress tracking
  • Quality assurance
  • Minimal user intervention required

​
/parallel - Parallel Execution Mode

Execute tasks using parallel processing capabilities. 
/parallel "create full-stack authentication system"

​
Agent Factory and Blueprints

​
/factory - Agent Factory Dashboard

Access the agent factory for creating and managing custom agents. 
/factory
Factory Features:
  • Agent creation interface
  • Blueprint management
  • Agent configuration tools

​
/create-agent - Create Custom Agent

Create specialized agents for specific domains or tasks. 
# Create domain-specific agent
/create-agent "React Testing Expert"
/create-agent "Database Optimization Specialist"  
/create-agent "API Security Analyst"
Agent Creation Process:
  1. Define agent specialization
  2. Configure capabilities and tools
  3. Set operational parameters
  4. Deploy and activate agent

​
/launch-agent - Launch from Blueprint

Deploy agents from pre-configured blueprints. 
# Launch from saved blueprint
/launch-agent blueprint-id "implement user authentication"

# Launch with specific task
/launch-agent react-testing-blueprint "create comprehensive test suite"

​
Blueprint Management

​
/blueprints - List All Blueprints

Display all available agent blueprints. 
/blueprints
Blueprint Information:
  • Blueprint ID and name
  • Agent specialization
  • Creation date
  • Usage statistics

​
/blueprint - View Specific Blueprint

View details of a specific blueprint. 
/blueprint [blueprint-id]

​
/delete-blueprint - Remove Blueprint

Delete a specific agent blueprint. 
/delete-blueprint [blueprint-id]

​
/export-blueprint - Export Blueprint

Export blueprint configuration for sharing or backup. 
/export-blueprint [blueprint-id]

​
/import-blueprint - Import Blueprint

Import a blueprint configuration from file. 
/import-blueprint [file-path]

​
/search-blueprints - Search Blueprints

Search blueprints by name, specialization, or capabilities. 
/search-blueprints "react testing"
/search-blueprints "backend optimization"

​
Context and Streaming

​
/context - Workspace Context Management

Manage workspace context for focused agent operations. 
# Show current context
/context

# Set specific paths
/context src/components/ src/hooks/

# Set context with multiple directories
/context src/ tests/ docs/
Context Features:
  • Focused codebase analysis
  • Improved agent performance
  • Targeted operations
  • Reduced processing overhead

​
/stream - Agent Activity Streaming

Monitor live agent activity and execution streams. 
# View agent stream dashboard
/stream

# Monitor specific operations
/stream [operation-id]
Stream Features:
  • Real-time agent activity monitoring
  • Progress tracking
  • Error detection and reporting
  • Performance metrics

​
Agent System Architecture

​
Registered Agents

NikCLI uses a focused approach with two primary registered agents:
  • Universal Agent
  • VM Agent
Full-Stack Development Specialist64 Core Capabilities:
  • Frontend: React, Next.js, TypeScript, JavaScript, HTML, CSS
  • Backend: Node.js, API development, database operations
  • DevOps: CI/CD, Docker, deployment, monitoring
  • Testing: Unit tests, integration tests, e2e testing
  • Code Quality: Review, refactoring, optimization
Operational Mode:
  • Fully autonomous operation
  • Maximum 3 concurrent tasks
  • Complete file system access (security-restricted)
  • Command execution with approval system integration

​
Agent Communication

Agents communicate through:
  • Event Bus: Inter-agent communication
  • Task Queue: Work distribution
  • Context Sharing: Workspace information
  • Progress Reporting: Status updates

​
Best Practices

​
Agent Selection Guidelines

Universal Agent

Use for:
  • General development tasks
  • Full-stack implementations
  • Code optimization and review
  • Multi-technology projects

VM Agent

Use for:
  • Repository analysis
  • Isolated development
  • Security-sensitive operations
  • Environment setup

​
Performance Optimization

  • Context Management: Set specific workspace contexts to improve performance
  • Task Scoping: Define clear, focused tasks for better results
  • Resource Monitoring: Use /stream to monitor agent activity
  • Blueprint Reuse: Create and reuse blueprints for common tasks

​
Troubleshooting

​
Common Issues

Problem: Agent fails to complete tasks or becomes unresponsiveSolutions:
# Check agent status
/agents

# View debug information
/debug

# Clear context and retry
/clear
/agent universal-agent "retry task"
Problem: Agent produces suboptimal resultsSolutions:
# Set specific context
/context src/components/

# Use more detailed task description
/agent universal-agent "create React login component with TypeScript, 
form validation, error handling, and responsive design"

# Try autonomous mode for complex tasks
/auto "comprehensive task description with requirements"
Problem: Blueprint creation or deployment failsSolutions:
# List available blueprints
/blueprints

# Search for similar blueprints
/search-blueprints "relevant keywords"

# Create new blueprint with specific parameters
/create-agent "Descriptive Agent Name"

​
Next Steps

Mode Control

Learn about operating modes and transitions

File Operations

Master file system operations with agents

Terminal Operations

Execute commands safely through agents

Universal Agent Guide

Deep dive into Universal Agent capabilities
Start with /agents to see available agents, then use /agent universal-agent "task" for most development tasks. For complex projects, try /auto "description" to enable autonomous execution.