Skip to content

concepts.md

Latest commit

 

History

History
197 lines (133 loc) · 6.23 KB

concepts.md

File metadata and controls

197 lines (133 loc) · 6.23 KB
Raw

Core Concepts

Understanding KubeRocketAI: AI-as-Code framework for enterprise development teams.

The Problem We Solve

Before KubeRocketAI

You've mastered individual AI agents, but scaling is broken:

Developer A: "I have the perfect PM agent that writes amazing PRDs"
Developer B: "Cool, can you share it?"
Developer A: "Uh... I'll copy-paste the prompt in Slack?"

Common scaling problems:

  • 🔄 Context Switching Overhead: Jumping between IDE and ChatGPT breaks development flow
  • 🎯 Project Context Blindness: AI agents don't understand your architecture, standards, or dependencies
  • 📋 Inconsistent Agent Management: Every developer has different setups, no organizational standards
  • 🔍 No Transparency: AI-generated changes aren't traceable or auditable
  • ⚠️ Configuration Drift: What worked last week doesn't work today, but you can't track why

The Real Cost

Based on our research with enterprise teams:

  • 5-10 minutes daily fixing AI-generated code that doesn't match project standards
  • 15% increase in code review cycles due to inconsistent AI outputs
  • 40% of organizations lack AI change tracking for compliance

Our Solution: AI-as-Code

Apply Pipeline-as-Code to AI Agents

Just like you've solved CI/CD pipeline management:

# Old Way: Manual server setup
"copy-paste from docs, hope for consistency..."

# New Way: Pipeline-as-Code
krci-ai install --ide=cursor

KubeRocketAI applies the same principle to AI agents:

# Old Way: Manual AI configuration
"Copy prompts, paste in ChatGPT, hope for consistency..."

# New Way: AI-as-Code
krci-ai install --ide=cursor

Core Concepts

1. Agent-as-Code

Agents are version-controlled YAML files that live alongside your code:

# .krci-ai/agents/pm.yaml
agent:
  identity:
    name: "Peter Manager"
    role: "Senior Product Manager"
    goal: "Drive product success through strategic planning"

  commands:
    create-prd: "Create comprehensive PRD by executing task create-prd"

  tasks:
    - ./.krci-ai/tasks/create-prd.md

Benefits:

  • ✅ Version Control: Track what works, roll back when things break
  • ✅ Team Sharing: Git workflows for agent distribution
  • ✅ Quality Assurance: Validate agents before deployment

2. Project Context Awareness

Agents understand YOUR project:

# Traditional AI
"Write a React component"
→ Generic code that doesn't match your patterns

# KubeRocketAI
"Write a React component" + Project Context
→ Code that follows your architecture, uses your dependencies, matches your standards

How it works:

  • Agents read your project's coding standards
  • Templates ensure consistent output formatting
  • Data files provide architectural patterns and best practices

3. Multi-Platform Deployment

One agent definition, multiple deployment targets:

# IDE Integration: Focused development context
krci-ai install --ide=cursor

# Web Chat: Full context for strategic discussions
krci-ai bundle --all --output strategy-session.md

Deployment modes:

  • IDE Mode: Lightweight, coding-focused context for daily development
  • Web Chat Mode: Complete project context for brainstorming and planning

4. Framework Structure

Six role-based agents covering the complete SDLC:

Agent Role Purpose
PM Product Manager Strategy, requirements, roadmap
PO Product Owner Backlog management, story creation
BA Business Analyst Requirements analysis, workflows
Architect System Architect Technical design, architecture decisions
Developer Software Engineer Code implementation, reviews
QA Quality Engineer Testing, validation, quality metrics

Each agent has:

  • Tasks: Step-by-step procedures for common workflows
  • Templates: Consistent output formatting
  • Data: Reference materials and organizational standards

Value Propositions

For Individual Developers

  • Faster Context Switching: Agents live in your IDE, no platform jumping
  • Better Code Quality: AI understands your project standards and patterns
  • Consistent Outputs: Templates ensure professional, standardized deliverables

For Development Teams

  • Agent Reusability: Share proven configurations across projects
  • Quality Assurance: Validate agent configurations before team deployment
  • Version Control Integration: Track agent changes like infrastructure changes

for Organizations

  • Scalable AI Governance: Organizational standards applied consistently
  • Audit Trail: Complete transparency in AI-assisted development
  • Reduced Maintenance: Centralized agent management with local customization

Key Differentiators

vs. Individual AI Tools (ChatGPT, Claude)

  • ✅ Project Context: Agents understand your specific architecture and standards
  • ✅ IDE Integration: No context switching, native development workflow
  • ✅ Team Scaling: Share and version-control agent configurations

vs. AI-Enhanced IDEs (Cursor, GitHub Copilot)

  • ✅ Organizational Governance: Centralized standards with local customization
  • ✅ Complete SDLC Coverage: Beyond coding - planning, architecture, testing
  • ✅ Multi-Platform: Same agents work in IDE, CI/CD, and web chat tools

vs. Enterprise AI Platforms

  • ✅ Lightweight: CLI tool, not heavy SaaS platform
  • ⚡ Fast Adoption: 5-minute setup vs. 3-6 month implementations
  • 💰 Cost Effective: Open source vs. enterprise licensing

Success Metrics

How you know it's working:

  1. Agent Reusability: Deploy proven configurations across multiple projects
  2. Quality Assurance: Validate agent configurations before deployment
  3. Version Control Integration: Track what works, roll back when things break
  4. Platform Flexibility: Use same agents for IDE development, CI automation, brainstorming

Target outcomes:

  • 75% reduction in time fixing AI-generated code
  • 90% of new users productive within 15 minutes
  • Team-wide consistency in AI-assisted development

Next: Ready to understand how it all works? Check out Architecture or jump back to Quick Start to get hands-on.