Your First Solution Play

This guide walks you through scaffolding, exploring, deploying, and evaluating Play 01: Enterprise RAG Q&A — the most popular FrootAI solution play.

What You’ll Build

A production-ready RAG (Retrieval-Augmented Generation) pipeline that deploys:

Service	Purpose	SKU (Dev)
Azure OpenAI	LLM inference (GPT-4o)	S0, 30K TPM
Azure AI Search	Vector + hybrid retrieval	Basic
Azure Container Apps	Application hosting	Consumption
Azure Key Vault	Secrets management	Standard

Prerequisites

FrootAI installed (Installation)
Azure CLI authenticated (az login)
Azure subscription with Contributor role

Step 1 — Choose Your Play

Browse available plays to find the right fit:

CLI


# List all 100 solution plays
npx frootai plays list
 
# Get details for Play 01
npx frootai plays get 01

JavaScript


import { FrootAI } from 'frootai';
 
const fai = new FrootAI();
 
// List all plays
const plays = fai.plays.list();
console.log(`${plays.length} plays available`);
 
// Get Play 01 details
const play = fai.plays.get('01');
console.log(play.name, play.description);

Python


from frootai import FrootAI
 
fai = FrootAI()
 
# List all plays
plays = fai.plays.list()
print(f"{len(plays)} plays available")
 
# Get Play 01 details
play = fai.plays.get("01")
print(play.name, play.description)

💡

Use the Configurator for an interactive recommendation wizard that matches your requirements to the best play.

Step 2 — Initialize the DevKit

Scaffold the complete play structure:


npx frootai init-devkit 01
cd solution-plays/01-enterprise-rag

This creates the full four-kit structure:


01-enterprise-rag/
├── .github/                        # DevKit — Copilot brain
│   ├── copilot-instructions.md     #   Always-on solution context (under 150 lines)
│   ├── agents/
│   │   ├── builder.agent.md        #   Builds the RAG pipeline
│   │   ├── reviewer.agent.md       #   Reviews security + quality
│   │   └── tuner.agent.md          #   Validates config + thresholds
│   ├── instructions/
│   │   └── rag-patterns.instructions.md
│   ├── skills/
│   │   └── rag-indexer/SKILL.md    #   Step-by-step indexing procedure
│   └── hooks/
│       └── guardrails.json         #   SessionStart event guardrails
├── config/                         # TuneKit — customer-tunable params
│   ├── openai.json                 #   Model, temperature, tokens
│   └── guardrails.json             #   Quality thresholds
├── spec/                           # SpecKit — architecture docs
│   └── 001-architecture.md
├── infra/                          # Infrastructure — Azure Bicep
│   ├── main.bicep
│   └── parameters.json
├── evaluation/                     # Evaluation pipeline
│   └── test-set.jsonl
└── fai-manifest.json               # FAI Protocol wiring

Step 3 — Explore the Generated Files

The Manifest (`fai-manifest.json`)

The manifest is the play’s DNA — it wires everything together:


{
  "play": "01-enterprise-rag",
  "version": "1.0.0",
  "context": {
    "knowledge": ["R2-RAG-Architecture", "O4-Azure-AI-Services"],
    "waf": ["security", "reliability", "cost-optimization"]
  },
  "primitives": {
    "agents": ["./.github/agents/builder.agent.md"],
    "instructions": ["./.github/copilot-instructions.md"],
    "skills": ["./.github/skills/rag-indexer"],
    "hooks": ["../../hooks/frootai-secrets-scanner/"],
    "guardrails": {
      "groundedness": 0.95,
      "coherence": 0.90,
      "safety": 0,
      "costPerQuery": 0.02
    }
  }
}

TuneKit Config (`config/openai.json`)

Customer-tunable AI parameters — adjust these for your use case:


{
  "model": "gpt-4o",
  "temperature": 0.3,
  "max_tokens": 4096,
  "top_p": 0.95,
  "fallback_model": "gpt-4o-mini"
}

Step 4 — Deploy to Azure

Create the Resource Group


PLAY="01-enterprise-rag"
ENV="dev"
RG="rg-frootai-${PLAY}-${ENV}"
 
az group create --name "$RG" --location eastus2 \
  --tags play="$PLAY" environment="$ENV" managed-by="frootai"

Preview Changes

Always preview before deploying:


az deployment group what-if \
  --resource-group "$RG" \
  --template-file infra/main.bicep \
  --parameters infra/parameters.json \
  --mode Incremental

Deploy Infrastructure


az deployment group create \
  --resource-group "$RG" \
  --template-file infra/main.bicep \
  --parameters infra/parameters.json \
  --name "frootai-${PLAY}-$(date +%Y%m%d)" \
  --mode Incremental

⚠️

Ensure you have sufficient Azure OpenAI quota in your target region. Request quota increases in the Azure Portal if needed.

Step 5 — Run Evaluation

Validate Primitive Wiring


# Schema + naming + frontmatter validation
npm run validate:primitives
 
# Load play in the FAI Engine
node engine/index.js fai-manifest.json --status

Run Quality Gates


# Evaluate against guardrail thresholds
node engine/index.js fai-manifest.json --eval

All guardrails defined in fai-manifest.json must pass:

Metric	Threshold	What It Measures
Groundedness	≥ 0.95	Responses supported by retrieved sources
Coherence	≥ 0.90	Logical consistency and readability
Safety	0 violations	No harmful content generated
Cost per Query	≤ $0.02	Token usage within budget

Post-Deployment Health Check


APP_URL="https://your-container-app.azurecontainerapps.io"
 
# Health check
curl -sf "${APP_URL}/health" | jq .
 
# Smoke test
curl -sf -X POST "${APP_URL}/api/chat" \
  -H "Content-Type: application/json" \
  -d '{"query":"What is the onboarding process?"}' | jq '.answer'

What’s Next?

FAI Protocol — understand how the manifest wires primitives
Solution Plays — explore all 100 plays
Well-Architected Framework — the 6 WAF pillars
How to Contribute — create your own play