Top AI Product

humanlayer’s 12-factor-agents repo climbed back onto GitHub trending today with 359 daily stars. It’s a manifesto plus reference material for building LLM-powered software that’s actually production-grade — modeled on Heroku’s original 12-Factor App methodology.

## The core thesis

“Agents, at least the good ones, don’t follow the ‘here’s your prompt, here’s a bag of tools, loop until you hit the goal’ pattern. Rather, they are comprised of mostly just software.” The author (Dex, founder of HumanLayer) interviewed roughly 100 SaaS builders making their existing products more agentic, then distilled the patterns into 12 factors.

## Highlight factors

– Tools are structured outputs — clean separation between LLM decision-making and deterministic code execution
– Own your context window — context engineering is the whole job; treat LLMs as stateless input-to-output functions
– Human feedback integration — well-scoped tasks let you fold live human review into workflow steps without context-error loops

## Why it matters

Agent frameworks (LangChain, CrewAI, LangGraph, smolagents, griptape) keep shipping and developers keep getting burned at the production line. The 12-factor framing reframes the problem: stop treating agents as magic loops, treat them as software with mostly-determined logic and a sprinkle of LLM judgment. The repo is becoming the de-facto reading list for serious agent builders.