_ ____ _ _ ___ __ _ ___ ___ ___ / \ / ___|| || | / _ \ / _| |/ _ \ / _ \ / _ \ / _ \ \___ \| || |_| | | | |_| | | | | | | | | | | / ___ \ ___) |__ _| |_| | _| | |_| | |_| | |_| | /_/ \_\____/ |_| \___/|_| |_|\___/ \___/ \___/
The Open-Source AI Agent Platform — Build, Deploy, and Orchestrate Intelligent Agents at Scale
Introduction • Features • Architecture • Tech Stack • Getting Started • Usage • Structure • Roadmap
AgentForge is a comprehensive, open-source platform for building, deploying, and orchestrating AI agents. Think of it as LangChain's flexibility meets CrewAI's team-based orchestration with n8n's visual workflow builder — all wrapped in a production-grade, full-stack application.
Unlike fragmented solutions where you stitch together LLM providers, memory systems, tool registries, and UIs, AgentForge provides a unified platform with:
- A visual agent builder for designing complex agent behaviors without coding
- Multi-LLM orchestration across OpenAI, Anthropic, Google, and local models
- Team-based collaboration where agents delegate, debate, and coordinate
- Extensible tool ecosystem with a built-in marketplace
- Enterprise-grade security with role-based access control and audit logging
Whether you're building a customer support bot, an automated code reviewer, a research assistant, or a complex multi-agent workflow, AgentForge gives you the foundation to go from idea to production in minutes.
- 🧠 Agent Creation Engine — Design agents with custom personas, goals, constraints, and knowledge bases
- 👥 Agent Teams — Form collaborative teams with role-based delegation (leader, researcher, executor, reviewer)
- 💾 Persistent Memory — Short-term (conversation) and long-term (vector store) memory with automatic RAG
- 🔧 Tool System — Extensible tool registry with 50+ built-in tools and MCP (Model Context Protocol) support
- 🛒 Agent Marketplace — Discover, install, and publish pre-built agents, tools, and templates
- 🔄 Visual Workflow Builder — Drag-and-drop workflow design with branching, loops, and conditions
- 🤖 Multi-LLM Support — OpenAI, Anthropic Claude, Google Gemini, Mistral, Groq, Ollama, and more
- 🐙 GitHub Integration — Code review automation, issue triage, PR analysis, and repo management
- 🌐 Browser Automation — Headless browser agent for web scraping, testing, and automation
- 💬 Slack Integration — Deploy agents directly to your Slack workspace
- 🔌 REST API — Comprehensive API for programmatic control of all platform features
- 📊 Real-time Dashboard — Monitor agent activity, performance metrics, and usage analytics
- 🔒 Enterprise Security — RBAC, SSO, audit logs, encryption at rest, and SOC 2 compliance ready
- 📦 Containerized Deployment — Docker Compose setup for one-command deployment
- 🧪 Testing Framework — Built-in agent testing sandbox with scenario simulation
- 🌍 Webhook Support — Trigger and receive webhooks for external system integration
AgentForge follows a modular, microservices-inspired architecture built on a monorepo structure with clear separation of concerns:
┌─────────────────────────────────────────────────────────────────────────────┐
│ CLIENT LAYER │
│ ┌─────────────┐ ┌─────────────┐ ┌─────────────┐ ┌────────────────────┐ │
│ │ Next.js 15 │ │ Agent CLI │ │ Mobile App │ │ External API │ │
│ │ (Web App) │ │ (Terminal) │ │ (Future) │ │ Clients │ │
│ └──────┬──────┘ └──────┬──────┘ └──────┬──────┘ └──────────┬─────────┘ │
└─────────┼────────────────┼────────────────┼─────────────────────┼───────────┘
│ │ │ │
┌─────────┼────────────────┼────────────────┼─────────────────────┼───────────┐
│ ▼ ▼ ▼ ▼ │
│ API GATEWAY / tRPC ROUTER │
│ ┌──────────────────────────────────────────────────────────────────────┐ │
│ │ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ │
│ │ │ Agents │ │ Teams │ │Workflows │ │ Memory │ │ Tools │ │ │
│ │ │ Service │ │ Service │ │ Service │ │ Service │ │ Service │ │ │
│ │ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ │ │
│ │ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ │ │
│ │ │ Auth │ │Marketplace│ │ Queue │ │Monitor │ │ Webhook │ │ │
│ │ │ Service │ │ Service │ │ Service │ │ Service │ │ Service │ │ │
│ │ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────┘ │ │
│ └──────────────────────────────────────────────────────────────────────┘ │
└─────────────────────────────┬───────────────────────────────────────────────┘
│
┌─────────────────────────────┼───────────────────────────────────────────────┐
│ ▼ │
│ DATA & INFRASTRUCTURE │
│ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────┐ ┌──────────────┐ │
│ │PostgreSQL│ │ Redis │ │ BullMQ │ │ S3/ │ │ Vector DB │ │
│ │(Primary) │ │ (Cache) │ │ (Queue) │ │ Local │ │ (pgvector) │ │
│ └──────────┘ └──────────┘ └──────────┘ └──────────┘ └──────────────┘ │
└─────────────────────────────────────────────────────────────────────────────┘
- Type Safety End-to-End — tRPC ensures typesafe communication from database to UI
- Event-Driven Architecture — BullMQ handles async agent execution, webhooks, and notifications
- Pluggable Providers — LLM providers, vector stores, and file storage are all swappable
- Stateless Services — Horizontal scaling ready with shared-nothing design
- Observability First — OpenTelemetry instrumentation across all services
| Layer | Technology | Purpose |
|---|---|---|
| Frontend | Next.js 15 (App Router), React 19, Tailwind CSS, Radix UI | Web application UI |
| Type Safety | TypeScript (Strict), tRPC, Zod | End-to-end type safety |
| Backend | Express.js, tRPC, Node.js 20+ | API and server logic |
| Database | PostgreSQL 15+, Drizzle ORM, pgvector | Primary data store & vector embeddings |
| Cache | Redis 7+ (Upstash/ioredis) | Session cache, rate limiting, pub/sub |
| Queue | BullMQ | Async job processing & scheduling |
| Auth | NextAuth.js v5, JWT, OAuth | Authentication & authorization |
| Storage | Local filesystem, S3-compatible (AWS, R2, MinIO) | File & asset storage |
| LLM Clients | AI SDK (Vercel), LangChain, OpenRouter | Multi-provider LLM integration |
| Monitoring | OpenTelemetry, Sentry, Pino | Logging, tracing, error tracking |
| Testing | Vitest, Playwright, MSW | Unit, E2E, and API mocking |
| CI/CD | GitHub Actions, Docker, Docker Compose | Build, test, deploy |
| Deployment | Docker, Vercel, Railway, Render, self-hosted | Platform deployment |
Screenshots will be available in the
/apps/web/public/screenshotsdirectory once generated.
| Dashboard | Agent Builder | Workflow Editor |
|---|---|---|
![]() |
![]() |
![]() |
| Chat Interface | Marketplace | Analytics |
![]() |
![]() |
![]() |
- Node.js 20.x or later (LTS recommended)
- PostgreSQL 15+ (with pgvector extension)
- Redis 7+
- Docker (optional, for containerized development)
- pnpm (recommended) or npm
git clone https://github.com/sudarshandev-llm/agent-forge-prototype.git
cd agent-forge-prototypepnpm installcp .env.example .env.localEdit .env.local with your configuration:
# Database
DATABASE_URL="postgresql://user:password@localhost:5432/agentforge"
# Redis
REDIS_URL="redis://localhost:6379"
# Auth
AUTH_SECRET="your-secret-key"
AUTH_GITHUB_ID="your-github-oauth-app-id"
AUTH_GITHUB_SECRET="your-github-oauth-app-secret"
# LLM API Keys
OPENAI_API_KEY="sk-..."
ANTHROPIC_API_KEY="sk-ant-..."
GOOGLE_API_KEY="..."pnpm db:migratepnpm db:seedpnpm devVisit http://localhost:3000 to access AgentForge.
import { AgentForge } from '@agentforge/core';
const forge = new AgentForge();
const agent = await forge.agents.create({
name: 'Code Reviewer',
persona: 'Senior software engineer with expertise in TypeScript and React',
model: 'gpt-4o',
tools: ['github', 'code-analysis', 'linting'],
memory: {
type: 'conversation',
ttl: 3600,
},
});
await agent.run('Review the latest PR in the main repository');Create visual workflows in the built-in editor, or define them programmatically:
const workflow = await forge.workflows.create({
name: 'Bug Triage Pipeline',
steps: [
{ agent: 'triage-agent', action: 'classify_issue' },
{ agent: 'research-agent', action: 'find_similar_bugs' },
{ agent: 'reviewer-agent', action: 'assess_severity' },
{ agent: 'assigner-agent', action: 'assign_developer' },
],
onComplete: {
webhook: 'https://hooks.slack.com/services/...',
},
});# Install an agent from the marketplace
npx agentforge marketplace install agent-research-assistant
# Publish your own agent
npx agentforge marketplace publishagentforge/
├── apps/
│ ├── web/ # Next.js 15 web application
│ │ ├── app/ # App Router pages & API routes
│ │ ├── components/ # Shared React components
│ │ ├── lib/ # Client-side utilities
│ │ └── public/ # Static assets & screenshots
│ ├── api/ # Express API server
│ │ ├── src/
│ │ │ ├── controllers/ # Route controllers
│ │ │ ├── services/ # Business logic
│ │ │ ├── middleware/ # Auth, rate limiting, validation
│ │ │ └── routes/ # API route definitions
│ │ └── test/ # API tests
│ └── worker/ # BullMQ worker for async processing
│ ├── src/
│ │ ├── jobs/ # Job processors
│ │ └── queues/ # Queue definitions
│ └── test/ # Worker tests
├── packages/
│ ├── core/ # Core agent engine & abstractions
│ │ ├── src/
│ │ │ ├── agents/ # Agent creation & management
│ │ │ ├── teams/ # Team orchestration & delegation
│ │ │ ├── memory/ # Memory providers (conversation, vector)
│ │ │ ├── tools/ # Tool registry & built-in tools
│ │ │ └── llm/ # LLM provider adapters
│ │ └── test/ # Core unit & integration tests
│ ├── database/ # Drizzle ORM schema & migrations
│ │ ├── schema/ # Database table definitions
│ │ └── migrations/ # Migration files
│ ├── api/ # tRPC routers & middleware
│ │ ├── routers/ # tRPC route definitions
│ │ └── middleware/ # Auth, rate limiting, validation
│ └── shared/ # Shared types, utilities, configs
│ ├── types/ # Shared TypeScript types
│ ├── utils/ # Shared utilities
│ └── config/ # Shared configuration
├── infra/
│ ├── docker/ # Docker & Docker Compose configs
│ └── github/ # GitHub-specific infrastructure configs
├── scripts/ # Build & maintenance scripts
├── docs/ # Documentation
├── assets/ # Logos, branding, media
├── .github/ # GitHub templates & workflows
├── turbo.json # Turborepo configuration
├── package.json # Root package configuration
└── tsconfig.base.json # Base TypeScript configuration
AgentForge exposes a comprehensive REST and tRPC API. Key endpoints:
| Endpoint | Method | Description |
|---|---|---|
/api/agents |
CRUD | Manage agents (create, read, update, delete) |
/api/agents/:id/run |
POST | Execute an agent with input |
/api/teams |
CRUD | Manage agent teams |
/api/workflows |
CRUD | Manage workflows |
/api/workflows/:id/execute |
POST | Trigger a workflow execution |
/api/tools |
GET | List available tools |
/api/marketplace/packages |
GET | Browse marketplace |
/api/auth/* |
Various | Authentication endpoints |
Full API documentation is available in the /docs directory.
We welcome contributions from the community! See our Contributing Guide for detailed information.
Quick steps:
- Fork the repository
- Create a feature branch (
git checkout -b feat/amazing-feature) - Commit your changes (
git commit -m "feat: add amazing feature") - Push to the branch (
git push origin feat/amazing-feature) - Open a Pull Request
Please read our Code of Conduct before contributing.
- Agent creation and management engine
- Multi-LLM provider support (OpenAI, Anthropic, Google, Groq)
- Visual workflow builder
- Tool system with MCP protocol support
- Agent marketplace MVP
- GitHub integration
- Docker deployment
- [🔄] Real-time agent collaboration and communication
- [🔄] Advanced team orchestration (debate, voting, consensus)
- [🔄] Enterprise SSO (SAML, OIDC, LDAP)
- [🔄] Audit logging and compliance reporting
- [🔄] Performance optimization and caching improvements
- [🔄] Agent analytics and cost tracking dashboard
- [📅] Mobile companion app (React Native)
- [📅] Custom model fine-tuning pipeline
- [📅] Agent-to-agent communication protocol
- [📅] Multi-tenant organization support with billing
- [📅] Advanced RAG with hybrid search
- [📅] Plugin SDK for third-party developers
- [📅] Agent monitoring and observability suite
- [📅] Auto-scaling and self-healing infrastructure
- [📅] Compliance certifications (SOC 2, GDPR)
- [📅] Enterprise support tier
- [📅] On-premise deployment option
- [📅] Community governance model
AgentForge is MIT licensed. See the LICENSE file for details.
Want to become a contributor? Check our Contributing Guide to get started.
- GitHub Issues — Report bugs or request features
- Documentation — See the
/docsdirectory - Email — support@agentforge.dev
AgentForge stands on the shoulders of giants. Special thanks to:
- LangChain — For pioneering the LLM application framework concept
- CrewAI — For inspiring agent team orchestration patterns
- n8n — For the visual workflow builder paradigm
- Next.js — For the incredible React framework
- tRPC — For end-to-end typesafe APIs
- Drizzle ORM — For the lightweight, type-safe ORM
- BullMQ — For robust job queue management
- Vercel AI SDK — For streamlined LLM integrations
- All Contributors — Every contributor who helps make this project better
When setting up this repository on GitHub, add these topics:
ai-agentsagent-frameworkllmnextjstypescriptopen-sourcemulti-agentworkflow-automationnodejsdocker
Made with ❤️ by the Team alum Team
AgentForge — Build the future of intelligent automation.





