---
title: "ChatDev: Communicative Agents for Software Development"
license: CC-BY-4.0
license_url: https://creativecommons.org/licenses/by/4.0/
license_note: 'paper is arXiv non-exclusive distribution license (verified on arXiv abs page 2026-11-12); this pointer note is our own words and is CC BY 4.0; we link to the paper rather than reproduce it'
source: https://iinventedlight.com/library/notes/2307-07924-chatdev.md
source_license: "arXiv non-exclusive distribution license"
attribution: 'Pointer by iinventedlight (AI-assisted). Source: Chen Qian, Wei Liu, Hongzhang Liu, and 11 coauthors, ChatDev: Communicative Agents for Software Development (arXiv non-exclusive distribution license). This note links to the paper rather than reproducing it.'
author: 'iinventedlight (AI-assisted)'
date: 2026-11-12
status: published
approved_by: "Cole Withington, 2026-11-12 14:44 PT, Librarian chat widget t51s316: \"Yes, publish all 100 second-list arXiv notes\""
checked_on: 2026-11-12
as_of: 2026-11-12
confidence: medium
untested: 'Pointer note only. Paper not reproduced. Abs page and license checked 2026-11-12.'
category: arxiv
topic_area: 'ai-research: agents'
tags: arxiv, agents, software-engineering, multi-agent, chatdev
arxiv_id: 2307.07924
alias: 2307-07924-chatdev
type: pointer
format: pointer
sources:
  - id: S1
    title: "ChatDev: Communicative Agents for Software Development"
    publisher: arXiv
    authors: "Chen Qian and 13 coauthors (listed on the arXiv abs page)"
    url: https://arxiv.org/abs/2307.07924
    version_read: "v5 (5 Jun 2024)"
    accessed: 2026-11-12
    license: "arXiv non-exclusive distribution license"
---
# ChatDev: Communicative Agents for Software Development

## Summary
[ChatDev: Communicative Agents for Software Development](https://arxiv.org/abs/2307.07924) (2023) organizes specialized language-model agents into a small software company that designs, codes, and tests programs through structured conversation rather than through a pipeline of separate deep-learning modules.

What it introduced: a chat-powered framework in which a chat chain guides what agents should say, and a communicative dehallucination method guides how they should say it. Agents contribute across design, coding, and testing, and solutions come from their multi-turn dialogue. The authors observe that natural language helps at the design stage, while talking in programming language helps with debugging.

What it led to: a widely copied multi-agent software-company pattern, alongside other role-playing agent systems. Live notes: [2308-00352-metagpt](https://iinventedlight.com/library/notes/2308-00352-metagpt/), [2304-03442-generative-agents](https://iinventedlight.com/library/notes/2304-03442-generative-agents/), and [2303-17760-camel](https://iinventedlight.com/library/notes/2303-17760-camel/). TheAgentCompany, a workplace-task benchmark, is a full note in this week's batch F.

Where to read it: https://arxiv.org/abs/2307.07924 (v5, last revised 5 Jun 2024). Accepted to ACL 2024. Code: https://github.com/OpenBMB/ChatDev.

The paper is posted on arXiv under the arXiv non-exclusive distribution license. This note links to it rather than reproducing it. The note itself is CC BY 4.0.

License: this note is CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). Pointer by iinventedlight (AI-assisted). Paper license: arXiv non-exclusive distribution; this note links to it and does not reproduce it.

---

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