chaitin/MonkeyCode

What it is

MonkeyCode (branded 百智办公 · MonkeyAI) is an open-source AI workspace for teams. Users describe a task in natural language, attach materials, and let the AI produce deliverables across three modes: office work (PPT, documents, spreadsheets, file organization), coding (design, implementation, fixes, tests) and design (posters, graphic and interaction design, short drama production) README lines 15–42.

The project's differentiator is the team layer: an admin console shares models, knowledge bases, skills, experts, MCP connectors and rules across members, tracks resource usage and billing, controls access permissions, audits usage, and reports team statistics README lines 43–55.

Quick start

The fastest route is the hosted demo. Download the desktop client for your platform — Windows x64, macOS Apple Silicon/Intel or Linux x64 AppImage — from the client download section, enter the demo server address https://demo.work.baizhi.cloud/, and log in with an email verification code README lines 72–79. To inspect team management features, open the demo admin console with the published demo credentials.

For your own server, the README offers a one-command install:

bash -c "$(curl -kfsSL https://release.monkeycode-ai.com/public/monkeyai-install.sh)"

The stated minimum is 1 CPU core and 1 GB RAM, adjustable by team size and workload README lines 56–71. Full installation, proxy/TLS, backup and upgrade steps, and source-based Docker Compose startup are covered in the deployment guide and the repository operations notes.

Constraints

• The install command pipes a remote script into bash; review it before running and check the deployment docs rather than treating it as an audited package.
• Self-hosting keeps team data on your own server, but prompts and content sent to external model or tool providers still leave your infrastructure; scope permissions and provider choices accordingly README lines 130–133.
• Clients are available for Windows, macOS and Linux; Android, iOS, Linux ARM64 and HarmonyOS desktop are marked as coming later README lines 90–107.
• The free tier caps teams at 10 seats and excludes cloud computers and technical support; paid plans are quote-based README lines 108–119.
• Code can be viewed, modified and redistributed under AGPL-3.0 README lines 134–137.

HotRepo's take

On the 2026-11-07 TypeScript daily list, MonkeyCode recorded 5,084 stars and 768 forks, with 22 stars gained that day and a best TypeScript daily position of 5. That single snapshot shows a substantial existing user base rather than a breakout pace. The notable engineering signal is that the repository ships backend, admin console, landing site and API docs as separate documented modules README lines 148–156, so teams evaluating self-hosting have concrete code to inspect before committing.