Play 01
Enterprise RAG Q&A
Production RAG with hybrid search, semantic reranking, and pre-tuned guardrails.
Build a production-grade Retrieval-Augmented Generation system. AI Search indexes your documents, GPT-4o generates grounded answers with citations, and Container Apps hosts the API. Pre-tuned with temp=0.1, hybrid 60/40 search, top-k=5, and semantic reranker for optimal quality.
Architecture Pattern
RAG: hybrid search, chunking, semantic reranking
Azure Services
DevKit (.github Agentic OS)
- agent.md — root orchestrator with builder→reviewer→tuner handoffs
- 3 agents — RAG Builder (gpt-4o), Reviewer (gpt-4o-mini), Tuner (gpt-4o-mini)
- 3 skills — deploy (106 lines), evaluate (153 lines), tune (167 lines)
- 4 prompts — /deploy, /test, /review, /evaluate with agent routing
- .vscode/mcp.json — FrootAI MCP with Azure key inputs + envFile
TuneKit (AI Config)
- config/openai.json — temp=0.1, seed=42, JSON schema
- config/search.json — hybrid 60/40, top-k=5, threshold=0.78
- config/chunking.json — 512 tokens, semantic, 10% overlap
- config/guardrails.json — safety, PII redaction
- evaluation/eval.py — Faithfulness >0.90, Relevance >0.85, Groundedness >0.95
Tuning Parameters
Machine evidence
Certified runtime lifecycle
Maturity is calculated from contiguous, content-bound evidence. Missing or expired evidence demotes automatically; catalog claims cannot promote a play.
Deterministic evidence runner
Status: loading. Runs deterministic bounded fixtures at the edge with zero model calls, Azure operations, or cloud mutation.
Enterprise gate
Checks both What-If and production admission using server-derived maturity. It never calls Azure or deploys resources.
Repo Intelligence
v1A no-clone, revision-pinned map for agents and humans. Observed evidence is separated from inferred flow so the output stays useful without pretending to be a full call graph.