Kaido - Your AI Ops Coach

Learn infrastructure operations through guided AI diagnosis. Not just answers — understanding.

You:   "nginx won't start, says port 80 already in use"

Kaido: Let me help you understand what's happening...

       THOUGHT: Need to identify what process is using port 80
       ACTION:  lsof -i :80 -P -n

       Here's what I found and why it matters:
       ┌─────────────────────────────────────────────────────────┐
       │ apache2 is running on port 80                          │
       │                                                         │
       │ lsof  → "list open files" (in Unix, network = file)    │
       │ -i :80 → filter by port 80                             │
       │ -P -n  → show port numbers, skip DNS lookup            │
       └─────────────────────────────────────────────────────────┘

       SOLUTION: Stop apache2 first, then start nginx

       Now you know: Two web servers can't share the same port.
       Next time, you can diagnose this yourself!

Why Kaido?

The Problem: Ops expertise takes years to build. AI tools that "just solve problems" don't help you grow.

Our Approach: Kaido shows its reasoning, explains every command, and teaches you why — so you become the expert.

Traditional AI	Kaido
Gives you the answer	Teaches you the concept
Black box	Transparent reasoning
You stay dependent	You become independent

Who Is This For?

• CS Students: First time using a terminal? Start here.
• Career Changers: Transitioning to DevOps? Learn the fundamentals.
• Frontend Developers: Want to understand the backend? We'll guide you.
• Junior SREs: Building real-world debugging skills.

Quick Start

Installation

One-liner (macOS & Linux)

curl -fsSL https://raw.githubusercontent.com/RikaiDev/kaido/main/install.sh | bash

From Source

cargo install --git https://github.com/RikaiDev/kaido

Initialize

kaido init

Upgrading

Self-update (v0.1.2+)

kaido update

From older versions or manual upgrade

curl -fsSL https://raw.githubusercontent.com/RikaiDev/kaido/main/install.sh | bash

Your First Session

kaido

Then just describe your problem in plain language:

> nginx returns 404 for /api endpoint
> docker containers can't talk to each other
> pod keeps crashing in kubernetes

Kaido will diagnose step-by-step, explaining each action along the way.

Features

Learn-As-You-Go

Every command comes with context:

ACTION: kubectl get pods -n production

┌─ What This Means ───────────────────────────────────────┐
│ kubectl    → Kubernetes command-line tool              │
│ get pods   → List running containers                   │
│ -n         → Namespace flag (like a folder)            │
│ production → The environment we're checking            │
└─────────────────────────────────────────────────────────┘

Transparent Reasoning (ReAct Pattern)

See how an expert thinks:

THOUGHT   → "Need to check if the service is running"
ACTION    → systemctl status nginx
OBSERVE   → "Service is failed, exit code 1"
REFLECT   → "Config error likely, should validate"
THOUGHT   → "Let me check the configuration"
ACTION    → nginx -t
SOLUTION  → Found syntax error on line 42

Safe Learning Environment

• Risk Classification: Commands labeled Low/Medium/High/Critical
• Confirmation Prompts: Dangerous commands require explicit approval
• Audit Trail: Every session logged for review

Privacy-First

• Ollama integration for local LLM inference
• Supports llama3.2, mistral, qwen2.5, and more
• All data stays on your machine
• Cloud AI optional (Gemini API)

Supported Tools

Domain	Tools
Containers	docker, docker-compose, kubectl
Web Servers	nginx, apache2
Network	lsof, netstat, iptables, ufw
Databases	MySQL, PostgreSQL
CMS	drush (Drupal)

Configuration

Run the interactive setup wizard:

kaido init

This teaches you about Cloud vs Local AI while configuring:

Option 1: Gemini API (Cloud)

# Fast, powerful, requires internet
export GEMINI_API_KEY="your_key_here"

Option 2: Ollama (Local)

# Private, offline-capable
brew install ollama        # or curl -fsSL https://ollama.ai/install.sh | sh
ollama serve               # start the server
ollama pull llama3.2       # download a model

Option 3: Both (Recommended)

• Gemini for speed, Ollama as private fallback
• kaido init auto-detects and configures both

Config file: ~/.kaido/config.toml

MCP Server

Kaido exposes 6 tools via Model Context Protocol for Claude Code integration:

{
  "mcpServers": {
    "kaido": {
      "command": "kaido-mcp",
      "args": []
    }
  }
}

Available Tools:

• kaido_diagnose - Diagnose ops issues with AI reasoning
• kaido_execute - Execute commands with risk assessment
• kaido_explain - Get educational explanations for commands
• kaido_get_context - Get system environment info
• kaido_list_tools - List supported ops tools
• kaido_check_risk - Check command risk level

Part of RikaiDev

Kaido is one of five tools in the RikaiDev ecosystem:

Tool	Purpose
Kaido	Learn Ops through guided diagnosis
Cortex	AI memory for coding assistants
inboxd	Unified inbox with AI processing
Toki	Automatic time tracking
Mimamori	Workplace communication guardian

All tools share a philosophy: AI as coach, not replacement.

Development

# Build
cargo build

# Test
cargo test

# Run
cargo run

Roadmap

See our GitHub Issues for planned features:

• ReAct reasoning loop
• Multi-tool support
• Risk classification
• Explain mode
• Ollama local LLM
• MCP server integration (6 tools for Claude Code)
• Learning tracker with session summaries
• Adaptive mentoring based on skill level
• AI-native shell with LLM-powered explanations
• Self-update command (kaido update)

License

MIT

---

Kaido (海道) — The sea route. Your path to mastering infrastructure.