BigHugger
GH Repository · humanlayer

12-factor-agents

What are the principles we can use to build LLM-powered software that is actually good enough to put in the hands of production customers?

stars
25,964
30-day movement
+4716/day
Related entries
60
Connections
2
makeTypeScript12-factor-agents12-factorragprompt-engineeringorchestrationmemoryframeworkllmscontext-windowaiagents

A repository from humanlayer that lays out twelve principles for building LLM-powered agent software reliable enough for production customers, styled after the classic Twelve-Factor App methodology. It is a TypeScript project and ships a CLAUDE.md file.

It gives you a structured, opinionated checklist of design principles when moving LLM agents from demos to production.

Use it to

  • Review your agent architecture against twelve production principles
  • Guide LLM application design decisions before building
  • Set up agent projects using the shipped CLAUDE.md
  • Educate a team on context-window, memory, and orchestration practices

For Developers building LLM agents for production use

Role
rag
Language
TypeScript
Licence
custom (licence file present)
Forks
1,970
Open issues
15
Last push
2025-09-21
topicsagentsllmproductionprinciplesorchestrationprompt-engineering