MetaGPT

Research summary

MetaGPT organizes software delivery as collaboration among role-specific agents rather than a single coding loop. Product, architecture, engineering and review responsibilities can be represented as separate roles that exchange requirements, designs, tasks and code. Its explicit software-company metaphor and artifact pipeline make it a useful reference for multi-stage AI software factories, while the underlying framework remains programmable Python rather than a fixed hosted product.

For engine design, the important boundary is between orchestration and individual model calls: the workflow owns role sequencing and artifact flow while each role performs a bounded reasoning task. The repository also exposes message routing, environment abstractions and reusable role/action primitives, making it possible to study how a software factory keeps intermediate requirements and designs explicit instead of relying on an ever-growing conversational context.

Repository summary

Stars

70,367

Open issues

13

Last push

2026-01-21

Commits, 90 days

0

Repository activity

Dormant · 28/100

Version

0.8.1

Recorded catalogue figures. View repository data and provenance →

Classification

Pricing & services

Paid services unknown

Whether the provider offers paid products or services has not been established.

Licence scope

Social & updates

• Xx.com/MetaGPT_↗
  Official
  Link details for X

  Last checked

  2026-10-01