A free, MIT-licensed course of 542 lessons that builds AI engineering concepts from scratch in Python, TypeScript, Rust and Julia.
victordibia/designing-multiagent-systems
Companion code for the book Designing Multi-Agent Systems, including PicoAgents, a multi-agent framework built from scratch for teaching.
- Stars
- 1,366
- Forks
- 285
- Open issues
- 8
- Last commit
- 2 months ago
What is it?
This repository accompanies the book Designing Multi-Agent Systems and contains PicoAgents, a multi-agent framework written from scratch so developers can see how agent loops, tool calling, memory, workflows, and orchestration work internally. The code is organized by book chapter, with runnable examples for each topic.
PicoAgents covers single-agent building blocks such as tools, memory, streaming, middleware, and structured output, along with workflows, round-robin and LLM-driven orchestration, plan-based orchestration in the style of Magentic One, a computer-use agent built on Playwright, and an evaluation framework with LLM-as-judge scoring. It supports OpenAI, Azure OpenAI, and Anthropic clients, and any OpenAI-compatible endpoint such as Ollama or vLLM.
The repo also includes a web UI launched with `picoagents ui` that auto-discovers agents and workflows, along with comparison examples for Microsoft Agent Framework, Google ADK, and LangGraph.
Key features
- 01Builds agents with tools, memory, streaming, and middleware from first principles
- 02Provides workflow and orchestration modules, including round-robin, LLM-driven, and plan-based patterns
- 03Includes a Playwright-based computer-use agent for browser automation
- 04Offers an evaluation framework with LLM-as-judge and reference-based checks
- 05Ships a web UI that auto-discovers agents and workflows
- 06Works with OpenAI, Azure OpenAI, Anthropic, and OpenAI-compatible local endpoints
Use cases
- Learning how multi-agent systems work internally before adopting a larger framework
- Comparing the same agent patterns across PicoAgents, LangGraph, and Google ADK
- Prototyping an orchestrated agent team with a web UI for debugging runs
- Building browser automation agents with a vision-capable model
Quick start
- 1
Clone the repository
git clone https://github.com/victordibia/designing-multiagent-systems.git - 2
Enter the PicoAgents directory
cd picoagents - 3
Install the package in editable mode
pip install -e . - 4
Set your OpenAI API key
export OPENAI_API_KEY="your-api-key-here"
Things to know
- Requires an OpenAI API key or another LLM provider key to run the examples.
- Built as a teaching framework, so it is not positioned as a production-ready library.
Topics
FAQ
- What is designing-multiagent-systems?
- Companion code for the book Designing Multi-Agent Systems, including PicoAgents, a multi-agent framework built from scratch for teaching. This repository accompanies the book Designing Multi-Agent Systems and contains PicoAgents, a multi-agent framework written from scratch so developers can see how agent loops, tool calling, memory, workflows, and orchestration work internally. The code is organized by book chapter, with runnable examples for each topic.
- Is designing-multiagent-systems free to use? What license does it have?
- designing-multiagent-systems is open source under the Apache-2.0 license. Check the license terms before using it in commercial products.
- Is designing-multiagent-systems still maintained?
- Updated within 6 months. The last commit to designing-multiagent-systems was on Aug 24, 2026.
- What can I use designing-multiagent-systems for?
- Learning how multi-agent systems work internally before adopting a larger framework Comparing the same agent patterns across PicoAgents, LangGraph, and Google ADK Prototyping an orchestrated agent team with a web UI for debugging runs Building browser automation agents with a vision-capable model
- How do I install designing-multiagent-systems?
- Clone the repository: git clone https://github.com/victordibia/designing-multiagent-systems.git Enter the PicoAgents directory: cd picoagents Install the package in editable mode: pip install -e . Set your OpenAI API key: export OPENAI_API_KEY="your-api-key-here"
- Where can I get designing-multiagent-systems?
- The source code and installation instructions are on GitHub: https://github.com/victordibia/designing-multiagent-systems
Summary generated by AI from the project's README. GitHub stats refresh daily.
Related projects
A teaching harness for coding agents that plans a course around you, builds lessons, and adapts exercises from your recorded progress.
AutoGPT is an open-source platform for building, running, and scheduling AI agents that carry out multi-step workflows from plain-English descriptions or a visual builder.
A Python framework and runtime for building, serving, and managing AI agent platforms, with a web UI for administration.
Open-source platform that turns any topic or document into an interactive classroom with AI teachers, slides, quizzes and simulations.
An open source MCP server that connects AI agents, IDEs and apps to databases with prebuilt tools or custom tool definitions.