---
title: 'MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework'
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is CC BY 4.0 (verified on arXiv abs page 2026-11-11; icon by-4.0.png, http://creativecommons.org/licenses/by/4.0/); note is our own summary'
source: https://iinventedlight.com/library/notes/2308-00352-metagpt.md
source_license: 'CC BY 4.0'
attribution: 'Summary by iinventedlight (AI-assisted). Source: Sirui Hong, Mingchen Zhuge, Jiaqi Chen, Xiawu Zheng, Yuheng Cheng, Ceyao Zhang, et al. (15 authors), MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework (CC BY 4.0).'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-11
status: published
approved_by: "Cole Withington, 2026-11-11 20:58 PT, Grok Bot chat widget t1717s7407: \"Publish all 100 including the LLMs Encode Clinical Knowledge note\""
checked_on: 2026-11-11
as_of: 2026-11-11
confidence: medium
untested: 'Not reproduced. Abs page and HTML key sections read 2026-11-11. Named code URLs checked; code not run.'
category: arxiv
topic_area: 'ai-research: agents'
tags: arxiv, agents, multi-agent
arxiv_id: 2308.00352
alias: 2308.00352-metagpt
sources:
  - id: S1
    title: 'MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework'
    publisher: arXiv
    authors: 'Sirui Hong, Mingchen Zhuge, Jiaqi Chen, Xiawu Zheng, Yuheng Cheng, Ceyao Zhang, et al. (15 authors)'
    url: https://arxiv.org/abs/2308.00352
    version_read: 'v7 (1 Nov 2024)'
    accessed: 2026-11-11
    license: 'CC BY 4.0'
---
# MetaGPT: Meta Programming for A Multi-Agent Collaborative Framework

## Summary
MetaGPT is a multi-agent framework that encodes Standardized Operating Procedures (SOPs) into prompt sequences and assigns roles along an assembly-line workflow. The paper says solutions to more complex tasks are complicated through logic inconsistencies due to cascading hallucinations caused by naively chaining LLMs, and that SOP sequences allow agents to verify intermediate results and reduce errors. On collaborative software-engineering benchmarks, they report more coherent solutions than previous chat-based multi-agent systems. [S1]

## Claims
- id: C1
  claim: 'MetaGPT encodes Standardized Operating Procedures into prompt sequences.'
  source: S1
  evidence: "MetaGPT encodes Standardized Operating Procedures (SOPs) into prompt sequences"
- id: C2
  claim: 'It assigns diverse roles to agents in an assembly-line paradigm.'
  source: S1
  evidence: "utilizes an assembly line paradigm to assign diverse roles to various agents"
- id: C3
  claim: 'The project is pointed to a public GitHub repository.'
  source: S1
  evidence: "Our project can be found at https://github.com/geekan/MetaGPT"

## Verification
- Code available: Paper: https://github.com/geekan/MetaGPT, which returned HTTP 200 and redirected to https://github.com/FoundationAgents/MetaGPT on 2026-11-11. Not run.
- Reproduced: not reproduced (2026-11-11). Code not run.
- Related work: Live note 2308-08155-autogen. This is the paper-verification note for MetaGPT itself.
- Limits: This note follows the v7 text last revised 1 Nov 2024. It does not restate later benchmark numbers from the live repo. The comparison is to earlier chat-based multi-agent systems named by the authors.
- License: CC BY 4.0 on the abs page (2026-11-11; icon by-4.0.png; http://creativecommons.org/licenses/by/4.0/). Submitted 1 Aug 2023 (v1); last revised 1 Nov 2024 (v7). Sources: [S1] https://arxiv.org/abs/2308.00352, v7 (1 Nov 2024). checked_on: 2026-11-11.

## When to apply
Use when looking at structured multi-agent software teams that encode human working procedures, rather than free-form chat between agents.

## Facts
- id: F1
  claim: 'MetaGPT encodes Standardized Operating Procedures into prompt sequences for multi-agent collaboration.'
  confidence: medium
  date: 2026-11-11
  sources: [S1]
  license: CC BY 4.0
- id: F2
  claim: 'The framework assigns diverse roles to agents in an assembly-line workflow.'
  confidence: medium
  date: 2026-11-11
  sources: [S1]
  license: CC BY 4.0
- id: F3
  claim: 'The authors report more coherent solutions than previous chat-based multi-agent systems on collaborative software-engineering benchmarks.'
  confidence: medium
  date: 2026-11-11
  sources: [S1]
  license: CC BY 4.0

## Sources
- [S1] arXiv 2308.00352, https://arxiv.org/abs/2308.00352. License: CC BY 4.0. v7 (1 Nov 2024). Accessed 2026-11-11.

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

---

Agents: [/agents.md](/agents.md) | [/.well-known/agent-library.json](/.well-known/agent-library.json).