Skip to main content

Documentation Index

Fetch the complete documentation index at: https://koreai-v2-home-nav.mintlify.app/llms.txt

Use this file to discover all available pages before exploring further.

Step-by-step tutorials, from creating your first agent to deploying a production-ready system. Each tutorial builds on the previous one. The tutorials use three example projects that introduce distinct capabilities:
ProjectTutorialsCapabilities Covered
Bean & Brew Greeter / Travel AssistantTutorial 1Agent definition, tools, tool types, tracing
Hotel Booking AgentTutorial 2Flows, data collection, branching, tool results
Retail Support SystemTutorial 3Supervisor routing, multi-agent context, knowledge bases, RAG
Safe AssistantTutorial 4Guardrails, constraints, evaluations, channels, monitoring
The tutorials follow a progressive learning path. Start with Build Your First Agent and work through them sequentially for the best experience.

1. Build Your First Agent

Create a coffee shop greeter agent, then extend it into a travel assistant that calls external APIs.You’ll learn:
  • Define agents using AGENT, GOAL, PERSONA, INSTRUCTIONS, and LIMITATIONS blocks
  • Add inline, HTTP, MCP, and sandbox tool contracts
  • Import shared tools from .tools.abl files
  • Test agents and read trace output in Studio
Prerequisites: An Agent Platform 2.0 account with at least one project.

2. Add Structured Steps to an Agent

Add a FLOW section to a hotel booking agent that guides users step-by-step through a reservation.You’ll learn:
  • Structure a conversation with named steps and THEN transitions
  • Collect typed user input with GATHER
  • Branch on user responses with ON_INPUT
  • Handle tool call results with ON_SUCCESS and ON_FAIL
Prerequisites: Completed Build Your First Agent.

3. Multi-Agent and Knowledge

Build a retail support system with a supervisor that routes to specialist agents, then add a knowledge base for RAG-powered answers.You’ll learn:
  • Create a SUPERVISOR with HANDOFF routing rules
  • Pass context between agents with CONTEXT and RETURN
  • Ingest documents into a knowledge base
  • Connect knowledge base search tools to agents
Prerequisites: Completed Add Structured Steps to an Agent.

4. Safety, Testing and Publishing

Add guardrails, run automated evaluations, and publish your agent to a Web channel with production monitoring.You’ll learn:
  • Configure input and output GUARDRAILS for PII and toxicity
  • Enforce business logic with CONSTRAINTS
  • Create test personas, scenarios, and evaluators
  • Publish to a channel and monitor sessions and errors in production
Prerequisites: Completed Build Your First Agent.