← All blueprints

Multi-Agent Debate Autogen

blueprint · 3 agents
50000★ stars

What it does

AutoGen's design-pattern notebook for peer debate: solvers each produce an answer, exchange them, and refine over multiple rounds, with an aggregator taking the majority at the end. The pattern for squeezing accuracy out of hard reasoning problems.

From the community

microsoft/autogenopen-source example · MIT (code) / CC-BY-4.0 (docs) · indexed by SocketCatview source ↗

Agent flow

1 · Solvers
Answer independently
no tools · reasoning
2 · Exchange
Read each other's answers
group chat
3 · Aggregator
Take the converged answer
no tools · reasoning
get the code
View the source on GitHub ↗

This blueprint lives in its authors' repository. SocketCat indexes it and links you to the real thing.

More in Core patterns

Browse all →

Supervisor Team

com.socketcat/supervisor-team
LangGraph

A supervisor decides who works next — a researcher who searches or a coder who edits the repo — and assembles the result when the work is done.

★ 21402 servers
⑂ 312

Deep Research

com.socketcat/deep-research
LangGraph

Splits a hard question into sub-questions, researches them in parallel to keep the wall-clock low, then synthesizes one brief with citations.

★ 17551 server
⑂ 254

Agentic RAG

com.socketcat/agentic-rag
LangGraph

Retrieval that doesn't trust its first search: it grades the results, rewrites the query when they're weak, and only answers once the context is good.

★ 16801 server
⑂ 241