Skip to main content
Scaffold structured LangGraph 1.0+ projects and keep them maintainable as they grow.
Act Operator is a CLI and project standard that keeps that drift from happening. It makes Acts and Casts explicit, and it provides AI Skills that keep architecture, development, and testing aligned as the project grows. If your LangGraph project:
  • Started clean, but became hard to reason about
  • Grew beyond a single graph and turned messy
  • Was difficult to test, refactor, or explain to another developer
Act Operator exists to solve exactly that.

​
Why Act Operator

LangGraph makes it easy to build agentic workflows. What it does not give you is a standard way to organize them as they grow. Act Operator provides:
  • A clear, modular structure for state, nodes, agents, and tools
  • A scalable way to manage multiple graphs as Cast packages
  • AI-native development via built-in Skills and explicit project context
  • Opinionated defaults that reduce architectural guesswork

​
30-second quick start

Requires Python 3.11+. 
uvx --from act-operator act new
Want the full setup flow? Start with Install and then go to Quickstart below.

​
How it works

  • act new scaffolds a standardized Act project
  • Each Cast is an independent graph package with explicit boundaries
  • CLAUDE.md files store design context for humans and AI tools
  • Skills guide architecture, implementation, and testing workflows

​
What is an Act

An Act is a structured, modular unit of a LangGraph application. Think of it as:
  • A bounded graph with a clear purpose
  • A package with explicit state, nodes, agents, and dependencies
  • A design artifact that can be reasoned about, tested, and refactored
  • A collaboration surface for both humans and AI agents
Multiple Acts (called Casts) can live together in a single repository, forming a maintainable monorepo for complex agentic systems.

​
Built for AI collaboration

Act Operator is designed for AI-assisted development from day one. Each project includes built-in Skills that allow AI tools to:
  • Understand your architecture before generating code
  • Ask the right questions instead of guessing
  • Follow consistent implementation and testing patterns
  • Evolve existing graphs without breaking structure

​
Next steps

Pick a path based on where you are:

Overview

Concepts, features, and target use cases.

Install

Install and setup details for Python projects.

Quickstart

Go from zero to a working cast fast.

Template architecture

Understand the generated structure and conventions.