Skip to content

parabala/Wegent

 
 

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

2,373 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

Wegent

An open-source, self-hostable platform for building and running AI agent teams.

English | 简体中文

CI License GitHub Issues

Wegent helps teams create and share AI agents that perform real work across chat, coding, knowledge, and automation. Manage team capabilities on the web, and use Wework when agents need to work directly with local projects and development environments.

Wegent product interface

What You Can Build

Scenario What Wegent provides
Team AI assistants A private chat entry point with shared models, knowledge, skills, group collaboration, and file handling
AI coding Change code, run tests, commit updates, and open pull requests in isolated or local environments
Knowledge assistants Parse and index documents, webpages, and enterprise data for grounded answers
Continuous automation Track information, analyze webpages, filter notifications, and publish feeds from schedules and events
Local and private-network execution Work with local code, CLIs, browsers, dedicated development environments, and intranet resources
Existing system integration Bring agents into applications and team tools through APIs, MCP, and IM bots

Choose How You Work

Use Wegent Web Use Wework
Create and share agents, models, knowledge bases, and automation Open local projects and let AI use files, terminals, CLIs, and development environments
Manage users, permissions, and execution devices Use local Codex, local models, and the local executor
Serve teams through the browser, APIs, and IM Focus on daily AI coding and local workflows

Wework can connect to a team deployment of Wegent to use shared models, cloud devices, and remote tasks while working locally.

Deploy Wegent · Download Wework

Why Choose Wegent

Capability Benefit
Reuse capabilities Combine models, knowledge, tools, and skills into agents that work across many tasks
Let bots collaborate Organize bots to divide research, analysis, coding, and review work
Run tasks in the right place Choose cloud, containers, local devices, or private environments based on where code and data live
Reach teams from every entry point Use the same agent capabilities from the web, Wework, APIs, and IM

Quick Start

Deploy Wegent Web

Prerequisite: Docker. Run Wegent in a single container with SQLite:

curl -fsSL https://raw.githubusercontent.com/wecode-ai/Wegent/main/install.sh | bash -s -- --standalone

After it starts:

  1. Open http://localhost:3000.
  2. Follow the setup flow to create the administrator password.
  3. Configure a model and API key in Settings.
  4. Choose a built-in agent and send your first message.

Common management commands:

wegent-standalone status
wegent-standalone logs
wegent-standalone restart
wegent-standalone stop

If the command is not in your current PATH, use ~/.local/bin/wegent-standalone.

Use Wework

Download Wework and open a local project to start AI coding. Wework includes local execution and can also connect to a team deployment of Wegent from Settings.

Download Wework Desktop

Wegent Web Deployment Options

Option Best for Start here
Standalone Personal trials and lightweight self-hosting Use the install command above
Standard Team deployments with MySQL, Redis, and dedicated services Installation Guide
Development Contributing and extending Wegent Development Setup

Standalone can also use host, container, or hybrid executor modes. See Standalone Mode for details.

Agent Model

See how Wegent organizes agents and tasks

Wegent manages capabilities, collaboration, and runtime context separately so they can be reused across tasks and environments.

Ghost (prompt + MCP + skills)
  + Shell (Chat / ClaudeCode / Agno / Dify)
  + Model
  = Bot

Multiple Bots + collaboration mode = Team (the user-facing Agent)
Team + Workspace = Task (a traceable execution)

Manage these resources through the UI, YAML, or APIs. See Core Concepts and the YAML Specification for details.

Architecture

View Wegent's technical components
graph TB
    User["User / API / IM"] --> Frontend["Wegent Web<br/>Next.js"]
    User --> Wework["Wework Desktop<br/>Tauri + React"]
    Frontend --> Backend["Backend<br/>FastAPI"]
    Wework -. "Optional cloud connection" .-> Backend
    Wework --> LocalWork["Local Codex / Files / Terminal"]

    Backend --> Database[("MySQL / SQLite")]
    Backend --> Redis[("Redis")]
    Backend --> ChatShell["Chat Shell"]
    Backend --> ExecutorManager["Executor Manager"]
    Backend --> KnowledgeRuntime["Knowledge Runtime"]

    ExecutorManager --> CloudExecutor["Cloud / Container Executor"]
    Backend <--> LocalExecutor["Local Executor"]
    KnowledgeRuntime --> VectorStore["Elasticsearch / Qdrant / Milvus"]
    KnowledgeRuntime --> DocConverter["Document Converter"]
Loading

Repository Map

Directory Responsibility
frontend/ Wegent Web product
backend/ REST API and core business logic
wework/ Tauri desktop workbench
executor/ Agent task execution environments
executor_manager/ Executor scheduling and orchestration
chat_shell/ Chat runtime
knowledge_runtime/ Knowledge retrieval services
knowledge_doc_converter/ Document parsing and conversion
shared/ Modules shared across services

Documentation

Get Involved

Bug reports, documentation improvements, code contributions, and new ways of using Wegent are all welcome.

Contributors

Thanks to everyone who helps Wegent grow.

qdaxb
Axb
Micro66
MicroLee
feifei325
Feifei
FicoHu
FicoHu
icycrystal4
Icycrystal4
cc-yafei
YaFei Liu
kissghosts
Yanhe
johnny0120
Johnny0120
parabala
Parabala
moqimoqidea
Moqimoqidea
yixiangxx
Yi Xiang
joyway1978
Joyway78
sunnights
Jake Zhang
cocowh
Birch
2561056571
Xuemin
fengkuizhi
Fengkuizhi
jnhu76
Jm.hu
kerwin612
Kerwin Bryant
maquan0927
Just Quan
RockysGit
RockysGit
DavidLeeUX
Kva
junbaor
Junbaor
fingki
Fingki
flyhope
李枨煊
hustfisher
fishermen
jolestar
Jolestar
code-wangdi
Code-wangdi
haosenwang1018
Sense_wang
LiDaiyan
Li Daiyan
sdadunderscoresdad
+7
qwertyerge
Erdawang
DeadLion
Jasper Zhong
rayzhang0603
Ray
RichardoMrMu
RichardoMu
Ged0
_
andrewzq777
Andrewzq777
ch15084
Ch15084
gdouyang
Gdouyang
graindt
Graindt
qingchengliu
Qingcheng
salt-hai
Salt-hai
wxcfox
Wxcfox

Made with ❤️ by WeCode-AI Team

About

An open-source AI-native operating system to define, organize, and run intelligent agent teams

Resources

License

Contributing

Security policy

Stars

0 stars

Watchers

0 watching

Forks

Packages

 
 
 

Contributors

Languages

  • Python 46.7%
  • TypeScript 42.1%
  • Rust 8.7%
  • Shell 1.5%
  • JavaScript 0.4%
  • CSS 0.3%
  • Other 0.3%