Introducing FrootAI — The UniFAIng Glue for the GenAI Ecosystem

The AI ecosystem is fragmented. MCP handles tool calling. A2A handles delegation. AG-UI handles rendering. But nothing wires them together — until now.

The Problem

Every AI team faces the same challenge: you have agents, prompts, skills, guardrails, tools, and evaluation pipelines — but they live in silos. Connecting them requires custom glue code, copy-pasted configurations, and tribal knowledge that doesn't scale.

Our Solution: The FAI Protocol

FrootAI introduces the FAI Protocol — a fai-manifest.json specification that declaratively wires context between all AI primitives. Think of it as package.json for AI architectures.

``json { "play": "01-enterprise-rag", "version": "2.0.0", "context": { "knowledge": ["rag-patterns.md", "azure-best-practices.md"], "waf": ["reliability", "security", "cost-optimization"] }, "primitives": { "agents": ["./agents/builder.agent.md"], "skills": ["./skills/index-documents/SKILL.md"], "hooks": ["./hooks/guardrails.json"] } } `

What's Included

- 100 Solution Plays — from Enterprise RAG to Edge AI to Swarm Orchestration - 830+ Primitives — agents, skills, instructions, hooks, plugins, workflows - 17 Learning Modules — the FROOT framework (Foundations, Reasoning, Orchestration, Operations, Transformation) - 5 Distribution Channels — VS Code, MCP Server, Python SDK, Docker, CLI

Built on Azure Well-Architected Framework

Every solution play aligns with all 6 WAF pillars: Reliability, Security, Cost Optimization, Operational Excellence, Performance Efficiency, and Responsible AI.

Get Started

`bash

MCP Server

What's Next

- FAI Engine — runtime that auto-wires primitives based on manifests - FAI Factory — CI/CD for AI primitives (lint, validate, publish) - FAI Marketplace — discover and share community primitives

Follow our journey on GitHub and LinkedIn.

FrootAI is open source under the MIT License. We welcome contributions from the community.