KubeRocketAI
>/quickstart>/architecture>/use-cases>/faq>/roadmap>/tutorials
Starred...
Pipeline-as-Code for AI

AI-as-Code for
Development Teams

Apply Pipeline-as-Code principles to AI agent management. Version-controlled, project-aware AI agents that understand your codebase.

terminal
$ |
Star on GitHub

Adoption at a glance

20+
Integrated Codebases
8+
Agile SDLC Roles
30+
Baseline Tasks
4
Supported IDEs
1
SDLC Framework

The Problem

  • ×Context switching between AI tools and IDEs
  • ×Inconsistent AI configurations across team members
  • ×Manual fixes for AI-generated code that doesn't fit project patterns
  • ×No version control for AI agent definitions

The Solution

  • ✓Declarative agent management with project-specific context
  • ✓Version-controlled Markdown definitions alongside code
  • ✓Universal compatibility across all AI-powered IDEs
  • ✓Built-in transparency and auditability for security teams

Key Features

Installation

Homebrew (Recommended)

brew tap KubeRocketCI/homebrew-tap brew install krci-ai

Having installation issues? Check our FAQ for troubleshooting help.

Quick Start

krci-ai install --ide=cursor
# Install framework with IDE integration
krci-ai validate
# Validate your agent configurations
krci-ai bundle --agent po
# Create PO context-aware bundle for web chat tools
krci-ai list agents
# List available agents
krci-ai install --all
# Install with all IDE integrations

How It Works

AI-as-Code methodology for development teams. Define AI agents in simple Markdown files that live with your codebase, understand your project context, and follow structured SDLC workflows.

No-Code Simplicity

Define agents using familiar Markdown + YAML format. No complex coding required—just structured documentation that agents understand.

Human-Centered

Agents assist, don't replace developers. Human-in-the-loop design ensures you stay in control of all decisions.

Version-Controlled

Agents live in your codebase, understand your patterns, and evolve with your project through Git version control.

Structured SDLC Workflow

IdeaConcept
BriefVision
PRDRequirements
EpicFeatures
StoryUser Needs
CodeImplementation

8+ specialized AI agents guide you through proven agile workflows, ensuring consistent quality from initial idea to delivered code.

IDE Integration

Native workflow in your editor

Cross-IDE compatible (Cursor, Claude Code, VS Code, JetBrains)
No plugins required—agents read Markdown files directly
Context-aware: agents understand your project structure
krci-ai install --ide=cursor

Web Chat Bundles

Portable context for any AI tool

Works with ChatGPT, Claude Web, Gemini Pro
Paste-ready bundles with complete project context
Validated packages ensure reliable AI interactions
krci-ai bundle --agent po

Context Engineering vs Static Prompting

Instead of static prompts that ignore your project, KubeRocketAI agents use dynamic context engineering—reading your codebase structure, existing patterns, and simple validation rules to provide intelligent, project-aware assistance that reduces token usage and improves accuracy.

Frequently Asked Questions

KubeRocketAI installation is designed to be completed in under 3 minutes:

1. Install CLI (30 seconds): Download and install the krci-ai CLI tool

2. Install Framework (20 seconds): Run krci-ai install to set up local agents

3. IDE Integration (automatic): Your AI-powered IDE automatically detects the framework

bash
# macOS installation
brew tap KubeRocketCI/homebrew-tap
brew install krci-ai

# Install framework with IDE integration
krci-ai install --ide=claude
# Done! Your IDE now has access to SDLC agents

For detailed installation steps and troubleshooting, see our Quick Start Guide.

KubeRocketAI works with any AI-powered IDE through automated configuration generation:

Officially Supported:

  • •Cursor - Generates .cursor/rules/*.mdc files for native integration
  • •Claude - Creates .claude/commands/*.md files for web-based IDE
  • •VS Code - Configures GitHub Copilot Chat integration via .vscode/
  • •Windsurf - Generates .windsurf/rules/*.md files for Cascade interface

    • Installation Examples:

    bash
    krci-ai install --ide=cursor   # Cursor IDE integration
krci-ai install --ide=claude   # Claude IDE integration
krci-ai install --ide=vscode   # GitHub Copilot integration
krci-ai install --ide=windsurf # Windsurf IDE integration
krci-ai install --ide=all      # All IDE integrations

    Universal Compatibility:

    The framework uses declarative AI-as-Code files that any AI IDE can read from the filesystem. No plugins or extensions required.

    See our complete IDE compatibility overview in the Supported IDEs section.

    Yes! KubeRocketAI enhances your existing AI workflow rather than replacing it:

    Compatibility with:

  • •GitHub Copilot Chat integration (via VS Code)
  • •Web chat tools (ChatGPT, Claude, Gemini) via bundle export
  • •Existing prompt libraries and custom workflows
  • •IDE-based AI assistants (Cursor, Claude Code, Windsurf)

    • How it works together:

  • •KubeRocketAI provides structured agent personas for SDLC tasks
  • •Your existing tools handle general coding assistance
  • •The framework maintains project context in version-controlled files
  • •Export bundles to web chat tools: krci-ai bundle --all --output project-context.md

    • Example Workflow:

    1. Use /pm agent in Claude Code for project planning

    2. Switch to GitHub Copilot for code completion

    3. Export framework context for web chat: krci-ai bundle --agent qa

    4. All context is maintained in your .krci-ai/ project files

    Learn how to set up integration with your existing tools in our Quick Start Guide.

    Even if your IDE isn't explicitly supported, KubeRocketAI provides several fallback options:

    Fallback Options:

    1. Bundle Export: Generate complete framework bundles for any AI tool

    bash
    krci-ai bundle --all --output my-framework.md
   krci-ai bundle --agent pm,architect --output targeted.md

    2. Direct File Access: Reference agent files from .krci-ai/ directory

  • • - Agents: .krci-ai/agents/*.yaml
  • • - Tasks: .krci-ai/tasks/*.md
  • • - Templates: .krci-ai/templates/*.md

    • 3. Manual Import: Copy generated bundles into your AI tool's prompt library

    Community Contributions:

  • •Request IDE Support: Submit feature requests for new IDE integrations
  • •Open Source: Submit compatibility reports for your IDE
  • •Documentation: Help us document setup for additional tools
  • •Testing: Join our community testing program

    • Universal Principles:

    The framework uses standard text files and markdown - any AI tool that can read files can use KubeRocketAI agents. The bundle system works with ChatGPT, Claude Web, Gemini Pro, and any other AI platform.

    Check our Quick Start Guide for alternative setup methods and supported IDE sections for community contributions.

    View All FAQ

    Ready to Transform Your AI Workflow?

    Join the next evolution of DevOps practices. Start managing your AI agents like infrastructure.Questions about implementation or adoption? View our FAQ for detailed answers.
    Get Started NowView on GitHub