---
source: https://iinventedlight.com/library/notes/ao-handoff-role-based-multi-agent.md
title: "Role-based multi-agent setups: SOPs and conversations"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
source_license: "CC BY 4.0; restated in our own words"
attribution: "Summary by iinventedlight (AI-assisted). Sources: arXiv, [2308.00352] MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework, CC BY 4.0; arXiv, [2308.08155] AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation, CC BY 4.0."
author: "iinventedlight (synthesis of cited sources); text drafted with AI assistance"
date: 2026-11-04
status: published
checked_on: 2026-11-04
as_of: 2026-11-04
confidence: medium
untested: "No experiment was rerun; only the fetched arXiv abstract pages were read. Results are as reported by the authors."
category: agent-operations
topic_area: "agent operations: multi-agent"
tags: agents, multi-agent, roles, workflows
sources:
  - id: S1
    title: "[2308.00352] MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework"
    publisher: arXiv
    url: https://arxiv.org/abs/2308.00352
    accessed: 2026-11-04
    license: "CC BY 4.0"
  - id: S2
    title: "[2308.08155] AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation"
    publisher: arXiv
    url: https://arxiv.org/abs/2308.08155
    accessed: 2026-11-04
    license: "CC BY 4.0"
---
# Role-based multi-agent setups: SOPs and conversations

## Short answer
Give each agent a clear role and a fixed procedure, and check work between steps. MetaGPT encodes standard operating procedures so errors do not cascade. AutoGen lets agents, tools and humans converse in set patterns. [S1][S2]

## Practice
Prefer an assembly line with checked handoffs over free chat between agents. [S1]

## Claims
- id: C1
  claim: Naive chaining causes cascading hallucinations.
  source: S1
  evidence: "cascading hallucinations caused by naively chaining LLMs"
- id: C2
  claim: MetaGPT encodes SOPs into prompts.
  source: S1
  evidence: "encodes Standardized Operating Procedures (SOPs) into prompt sequences"
- id: C3
  claim: MetaGPT uses an assembly line of roles.
  source: S1
  evidence: "utilizes an assembly line paradigm to assign diverse roles to various agents"
- id: C4
  claim: AutoGen agents combine LLMs, humans and tools.
  source: S2
  evidence: "employ combinations of LLMs, human inputs, and tools"

## Sources
- [S1] [2308.00352] MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework. arXiv. https://arxiv.org/abs/2308.00352 . License: CC BY 4.0. Accessed 2026-11-04.
- [S2] [2308.08155] AutoGen: Enabling Next-Gen LLM Applications via Multi-Agent Conversation. arXiv. https://arxiv.org/abs/2308.08155 . License: CC BY 4.0. Accessed 2026-11-04.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Summary by iinventedlight (AI-assisted).