MetaGPT
Multi-agent framework that role-plays a software company
Overview
MetaGPT assigns LLM agents software-team roles (PM, architect, engineer, QA) that collaborate via standard operating procedures to turn a one-line prompt into a PRD, system design, and working code. Now under the FoundationAgents org on GitHub, with 69k+ stars.
Pricing
Pricing shown for reference only. These figures reflect RECATOOLS research as of 11 Jul 2026 and may be out of date or incomplete. This is not financial or purchasing advice — always confirm the current price on the provider’s official website before making any decision.
Use cases
ASEAN Perspective
MetaGPT in Southeast Asia
ASEAN-region availability and pricing notes coming soon. Drop the editorial team a note via /contact/ if you can supply local context (Singapore/Malaysia/Indonesia/Thailand/Vietnam).
MetaGPT is the widely cited open-source multi-agent framework that assigns LLM 'agents' software-team roles — PM, architect, engineer, QA — to turn a one-line requirement into specs, designs, and code. It's now hosted under the FoundationAgents GitHub org (still MIT-licensed) with 69k+ stars, and as a demonstration of role-based agent orchestration it remains genuinely influential.
It's a research-grade framework, not a polished product: setup needs Python and API keys, output quality depends heavily on the underlying LLM, and generated projects typically need real human fixing before shipping. The team's commercial spin-off, MGX (rebranded Atoms), gets mixed reviews — some call it a genuine virtual dev team, others report it riddled with defects. Free, open-source, and globally accessible, but expect a tinkerer's experience over a turnkey app builder.
What people say
MetaGPT crossed 69,000 GitHub stars under its new home, the FoundationAgents org — the project moved off founder Sirui Hong's original geekan account at some point in the last two years, though the code and MIT license carried over intact.
Academic reception has stayed strong: the team's AFlow paper on automating agentic workflow generation landed an oral slot at ICLR 2025 (top 1.8% of submissions), and MetaGPT itself remains one of the most-cited multi-agent papers around. That citation count reflects real influence on how people think about role-based agent orchestration — PM, architect, engineer, and QA agents handing work to each other.
Hands-on feedback is rougher. On Product Hunt, reviews of the commercial spin-off (MGX, now rebranded Atoms) split hard: some users call the automated dev-team experience efficient and intuitive, others describe runs that were "nothing short of disastrous" and riddled with defects that blocked them from finishing anything. The consensus across both camps is that context handling, reliability, and integration depth still need work before this is dependable for production use.
For the open-source framework itself, that's the expected trade: a genuinely clever demonstration of coordinated agents that still needs a developer babysitting it, not a hands-off AI software company.
Summary of public user & expert reviews, compiled by RECATOOLS.
Notable facts
- MetaGPT's research paper was cited over 1,000 times within 6 months of publication, making it one of the fastest-cited papers in AI agent research history.
- The framework generates a complete PRD, system design, and implementation from a single natural language requirement — mimicking how a full software team works.
- MetaGPT demonstrated that role assignment (not just tool use) is a key factor in making multi-agent systems more capable than single agents.
Frequently asked questions
About this listing
This entry was compiled from publicly available data including MetaGPT's official website, press releases, documentation, and reputable third-party publications. RECATOOLS is not affiliated with MetaGPT unless explicitly stated.
Third-party AI tools update their pricing, features, availability, and policies frequently. Information here may be outdated by the time you read this — we make reasonable efforts to keep listings current, but cannot guarantee absolute accuracy.
For the latest details, please refer to MetaGPT directly →
Spotted something out of date? Suggest an update →
Alternatives to MetaGPT
More in Agents & Automation
