PromptMash

A modern AI agent pipeline platform with a stateless server and client-side configuration.

▶️ Watch Demo Video

Architecture

• Frontend (React + IndexedDB) – Stores pipeline configs, agent settings, and user data.
• Backend (Go) – Handles temporary pipeline execution sessions only.
• Stateless Server – No persistent storage; all data stays client-side.

Features

• 🤖 Chain multiple AI agents
• 🔧 Visual pipeline builder
• 🔄 Real-time monitoring for debugging
• 🎯 Multi-provider support: OpenAI, Anthropic, Google AI, Cohere, Hugging Face

Quick Start

Prerequisites

• Go 1.24.3+ – Required for backend development
• Node.js 24.1.0+ – Required for frontend (React) development
• API keys – Needed to run pipelines with AI providers

Setup

# Set API keys of your choice
export OPENAI_API_KEY="..."
export ANTHROPIC_API_KEY="..."
export GOOGLE_API_KEY="..."
export COHERE_API_KEY="..."
export HUGGINGFACEHUB_API_TOKEN="..."

git clone https://github.com//PromptMash.git
cd PromptMash


# Run the automated setup script (installs dependencies, and starts services)
./setup.sh

Open http://localhost:3000 in your browser.

API

• POST /pipelines/execute – Run a pipeline and get the result
• POST /pipelines/execute/stream – Run with live SSE updates

See API docs for details.

Project Structure

PromptMash/
├── agents/        # AI agent implementations
├── dashboard/     # React frontend
├── orchestration/ # Pipeline logic
├── server/        # Go backend
├── shared/        # Utilities and constants
└── main.go        # Backend entry

Development

• Quick Setup: ./setup.sh - Automated setup for both frontend and backend
• Frontend: cd dashboard && npm install && npm run dev
• Backend: go mod tidy && go run main.go
• Tests: go test ./... (backend), npm test (frontend)

Contributing

1. Fork → feature branch → commit → push → Pull Request

License

GNU AGPL-3.0 © 2025 Alexsander Hamir