AI Agent Orchestration Starter Kit
Production-ready multi-agent AI kit with memory & HITL
Start new thread

AI Agent Orchestration Starter Kit - Production-ready multi-agent AI kit with memory & HITL

by•
Multi-agent AI orchestration with a supervisor that decomposes tasks and delegates to specialists — plus persistent memory (Redis + ChromaDB), human-in-the-loop approval for sensitive actions, and full OpenTelemetry observability with cost tracking and replay. Built and battle-tested on a real production system — including fixing a runaway loop that generated 5,000+ duplicate DB rows before it was caught. One-command Docker setup, ~2 minutes to running.

Add a comment

Replies

Best
Maker
📌
Hey Product Hunt! 👋 I built this after shipping a full multi-agent orchestration system (LangGraph, memory, human-in-the-loop approvals, full tracing) as a personal project — then realized most of the real pain was in the infrastructure, not the agent logic itself. Along the way I hit and fixed some real production bugs: a runaway approval loop that generated 5,000+ duplicate rows, a silent checkpoint failure, and a replay identifier mismatch. Those fixes are already baked into this kit, so you don't have to hit them yourself. This is for developers building AI agent products who don't want to spend 2-3 months on the orchestration/memory/observability layer before they can even start on their actual product. Would love feedback — especially from anyone who's built something similar and hit different walls!