foundrai 0.1.0

AI agent orchestration as an autonomous Agile startup team

🏢 FoundrAI

Your AI-Powered Founding Team

Autonomous AI agents working as an Agile startup team — with a live visual dashboard.

Getting Started · Documentation · Roadmap · Contributing

---

What is FoundrAI?

FoundrAI is an open-source platform that orchestrates multiple AI agents as an autonomous Agile team. Give it a goal like "Build a REST API for a todo app", and a team of specialized AI agents (Product Manager, Architect, Developer, QA) will self-organize into sprints to plan, build, test, and iterate — while you watch everything unfold on a real-time dashboard.

Think of it as a virtual startup team you can observe, guide, and learn from.

Why FoundrAI?

Feature	CrewAI	MetaGPT	ChatDev	AutoGen	FoundrAI
Visual Dashboard	❌ (Studio is commercial)	❌	Basic	Basic	✅ Real-time mission control
Agile Sprints	❌	❌ (single pass)	❌ (waterfall)	❌	✅ Plan → Execute → Review → Retro
Human-in-the-Loop	Basic	❌	❌	Basic	✅ Granular per-agent policies
Iterative Learning	❌	❌	❌	❌	✅ Agents improve across sprints
Model-Agnostic	Partial	Partial	❌	✅	✅ Different LLM per agent role
Open Source UI	❌	❌	❌	Partial	✅ Fully open source

Key Features

🎯 Goal → Execution Pipeline — Enter a high-level goal and watch agents decompose, plan, and execute it

📊 Live Sprint Board — Real-time Kanban showing what each agent is working on

💬 Agent Communication Feed — See every message, decision, and reasoning trace between agents

🎛️ Configurable Autonomy — Set per-agent approval policies (auto-approve code but review architecture decisions)

🔄 Sprint Retrospectives — Agents analyze and improve their process after each sprint

🔌 Model-Agnostic — Assign Claude for PM, GPT-4o for coding, a local model for QA

📈 Performance Analytics — Track token costs, agent performance, and sprint velocity

Screenshots

Dashboard

Main dashboard with project stats, quick actions, and empty state onboarding

Project Creation Wizard

3-step project wizard — configure name, team, and settings with form validation

Template Browser

Browse and search project templates with tag filtering

Settings — General

Configure default model, autonomy level, and budget limits with breadcrumb navigation

Settings — API Keys

Manage LLM provider API keys — add, test, and remove with masked display

Settings — Appearance

Theme selector with Dark, Light, and System options

404 Page

Clean error page with navigation back to dashboard

Getting Started

Prerequisites

• Python 3.11+
• Node.js 18+ & npm (for the web dashboard)
• Docker (for sandboxed code execution)
• An API key for at least one LLM provider (OpenAI, Anthropic, etc.)

Installation

# Clone the repo
git clone https://github.com/kyled7/foundrai.git
cd foundrai

# Install Python backend
pip install -e ".[dev]"

# Install frontend dependencies
cd frontend-v2
npm install

Quick Start

# Configure your LLM provider(s)
export OPENAI_API_KEY="sk-..."
# and/or
export ANTHROPIC_API_KEY="sk-ant-..."

# Start the web dashboard (recommended)
cd frontend-v2
npm run dev
# → Open http://localhost:5173

# Or use the CLI directly
foundrai sprint start "Build a REST API for a todo app with authentication"

Web Dashboard (v0.2)

The dashboard is the primary interface as of v0.2. From the UI you can:

• Create projects via the 3-step wizard (name → team → settings)
• Browse templates to quick-start common project types
• Configure settings — default model, autonomy level, budget, API keys, theme
• Monitor sprints — real-time agent feed, Kanban board, approvals (coming in v0.3)

CLI Commands

FoundrAI also provides a CLI for headless usage:

foundrai doctor          # Check system health and prerequisites
foundrai sprint start    # Start a sprint from a goal description
foundrai status          # Project status and monitoring
foundrai logs            # View agent logs
foundrai serve           # Launch the web dashboard on port 8420

Architecture

┌─────────────────────────────────────────────────────┐
│              Frontend (React + TypeScript)            │
│  Sprint Board │ Agent Feed │ Goal Tree │ Metrics     │
├─────────────────────────────────────────────────────┤
│              API Layer (FastAPI + WebSocket)          │
├─────────────────────────────────────────────────────┤
│           Orchestration (LangGraph + SprintEngine)   │
├─────────────────────────────────────────────────────┤
│              Agent Layer (LiteLLM + Tools)            │
├─────────────────────────────────────────────────────┤
│           Persistence (SQLite + ChromaDB)             │
└─────────────────────────────────────────────────────┘

See Architecture Documentation for full details.

Agent Roles

Agent	Role	What They Do
📋 PM	Product Manager	Decomposes goals, writes stories, prioritizes backlog
🏗️ Architect	System Architect	Designs architecture, reviews technical decisions
💻 Dev	Developer	Writes code, submits for review, fixes bugs
🧪 QA	QA Engineer	Tests code, reports bugs, validates fixes
🎨 Designer	UI/UX Designer	Creates mockups, defines user flows
🚀 DevOps	DevOps Engineer	CI/CD setup, deployment, monitoring

Configuration

# foundrai.yaml
team:
  product_manager:
    model: anthropic/claude-sonnet-4-20250514
    autonomy: notify
  architect:
    model: anthropic/claude-sonnet-4-20250514
    autonomy: require_approval
  developer:
    model: openai/gpt-4o
    autonomy: auto_approve
    conditions:
      require_approval_if:
        lines_changed: ">50"
  qa_engineer:
    model: openai/gpt-4o-mini
    autonomy: auto_approve

sprint:
  max_tasks_parallel: 3
  token_budget: 100000

Roadmap

• v0.1 ✅ Core agent engine + CLI (Foundation, Visual Layer, Agile Engine, Observability, Ecosystem)
• v0.2 ✅ UI-First Platform — Full web dashboard replacing CLI
  • v0.2.0 ✅ Frontend Foundation (React 19, TypeScript, Tailwind, TanStack Router/Query)
  • v0.2.1 ✅ Dashboard & Project Management (wizard, team config, templates)
  • v0.2.2 ✅ Sprint Command Center (realtime feed, WebSocket, approvals)
  • v0.2.3 ✅ Analytics & Insights (charts, cost tracking, sprint replay)
  • v0.2.4 ✅ Settings & Polish (settings page, error handling, accessibility)
• v0.3 🔨 Backend Integration — Connect frontend to live backend

See full roadmap for details.

Contributing

We'd love your help! FoundrAI is in early development and there are many ways to contribute:

• 🐛 Report bugs and suggest features via Issues
• 🔧 Submit PRs (see our Contributing Guide)
• 📖 Improve documentation
• 🧪 Write tests
• 🎨 Design UI components
• 🔌 Build plugins and integrations

License

MIT License — see LICENSE for details.

---

Built with ❤️ by the FoundrAI community

⭐ Star us on GitHub · 📖 Read the Docs · 💬 Join the Discussion