Chatbot Example

Simple LLM chatbot using the unified API.

Features

• ✅ YAML configuration
• ✅ Unified API (get() function)
• ✅ LLM chat with Ollama
• ✅ JSON response
• ✅ Validation with preflight checks

Run Locally

# From examples/chatbot directory
kdeps run workflow.yaml

# Or from root
kdeps run examples/chatbot/workflow.yaml

Test

curl -X POST http://localhost:16395/api/v1/chat \
  -H "Content-Type: application/json" \
  -d '{"q": "What is artificial intelligence?"}'

Response

{
  "data": {
    "answer": "Artificial intelligence (AI) is..."
  },
  "query": "What is artificial intelligence?"
}

Structure

chatbot/
├── workflow.yaml              # Main workflow configuration
└── resources/
    ├── llm.yaml              # LLM chat resource
    └── response.yaml         # API response resource

Key Concepts

Unified API

Uses get() for all data access:

# Get query parameter
prompt: "{{ get('q') }}"

# Get LLM response from previous resource
data: get('llmResource')

# Validation
validations:
  - get('q') != ''

Auto-Detection

get() automatically detects the data source:

• get('q') → Query parameter
• get('llmResource') → Resource output
• get('user_data') → Memory storage