Solution Plays Overview

Solution Plays are pre-built, tested, production-ready AI solution blueprints. Each play provides everything you need to go from zero to deployed: Copilot agents, configuration files, infrastructure-as-code, evaluation pipelines, and documentation — all wired together by the FAI Protocol.

What’s in a Play?

Every solution play follows the 4-kit structure:

solution-play-NN/
├── .github/                          ← DevKit: Copilot brain
│   ├── copilot-instructions.md       ←   under 150 lines, domain knowledge
│   ├── agents/                       ←   builder / reviewer / tuner
│   │   ├── builder.agent.md
│   │   ├── reviewer.agent.md
│   │   └── tuner.agent.md
│   ├── instructions/                 ←   WAF-aligned guidance
│   ├── skills/                       ←   150+ line action skills
│   └── hooks/                        ←   SessionStart guardrails
├── config/                           ← TuneKit: customer-tunable params
│   ├── openai.json                   ←   model, temperature, max_tokens
│   └── guardrails.json               ←   safety thresholds
├── infra/                            ← Infra: Azure Bicep (AVM modules)
│   ├── main.bicep
│   └── modules/
├── evaluation/                       ← Eval: quality pipeline
│   └── eval-config.json
└── spec/                             ← SpecKit: metadata + docs
    └── fai-manifest.json             ←   FAI Protocol wiring

Categories

Featured Plays

See the full Play Catalog for all 23+ plays.

How to Browse

How to Initialize

# Scaffold any play by number + name
npx frootai scaffold 01-enterprise-rag
 
# Scaffold with specific options
npx frootai scaffold 02-ai-landing-zone --region eastus2
 
# List available plays
npx frootai list plays

After scaffolding:

1. Review .github/copilot-instructions.md for domain context
2. Customize config/openai.json and config/guardrails.json
3. Deploy infrastructure with az deployment group create --template-file infra/main.bicep
4. Use @builder agent in Copilot Chat to implement

The Agent Triad

Every play includes three specialized agents that follow the Build → Review → Tune workflow:

@builder → implements code and architecture
    ↓
@reviewer → checks security, quality, WAF alignment
    ↓
@tuner → validates config, thresholds, production readiness

WAF Alignment

All plays are aligned to the 6 Well-Architected Framework pillars:

• Reliability — retry, circuit breaker, health checks, fallbacks
• Security — Managed Identity, Key Vault, RBAC, private endpoints
• Cost Optimization — model routing, token budgets, right-sizing
• Operational Excellence — CI/CD, IaC, observability, incident response
• Performance Efficiency — caching, streaming, async, model routing
• Responsible AI — content safety, groundedness, fairness, transparency

For detailed WAF patterns, see T3: Production Patterns and T2: Responsible AI.

Next Steps

• Browse the full Play Catalog
• Try Workshop: Build a RAG Pipeline (Play 01)
• Try Workshop: AI Landing Zone (Play 02)
• Learn the foundations in the Learning Hub