Codeg

_{English · 简体中文 · 繁體中文 · 日本語 · 한국어 · Español · Deutsch · Français · Português · العربية}

The multi-agent coding workspace.
Run every AI coding agent in one place — and let them work together.

Download Codeg · Documentation

🧩 One interface for every agent Fifteen agents built in, and any ACP agent can join — all rendered as the same structured conversation, not a terminal.	🔎 Sessions that travel Import, search, and resume the history every agent keeps on disk, then hand it to a different agent.
🤝 Agents that work together Delegate across agents with an @, or queue to-dos that run unattended in their own worktrees.	🌍 Wherever you work Desktop app, self-hosted server or Docker, iPhone, iPad, and Android — and Telegram, Lark, or WeChat.

💖 Sponsors

Compshare (UCloud)	Thanks to Compshare for sponsoring this project! Compshare is UCloud's AI cloud platform, offering cost-effective monthly and pay-as-you-go agent Plan subscriptions for Chinese models, starting at just ¥49/month. It also provides stable officially-proxied access to overseas models. Supports Claude Code, Codex, and API integrations. Enterprise-ready with high concurrency, 24/7 technical support, and self-service invoicing. Users who sign up via this link receive ¥5 in free platform credits!
合租巴士	Thanks to 合租巴士 for sponsoring this project! 合租巴士 is a reliable and efficient AI relay platform, offering highly stable relay for mainstream models such as Codex and Claude Code. Top-ups are credited at a transparent 1:1 ratio, with Codex rate subsidies as low as 0.08. Join the group via the official website to get $5 in trial credit.
LQ router	Thanks to LQ router 中转站 for sponsoring this project! LQ router is a professional, enterprise-grade AI relay service that gives businesses and individual developers stable, efficient, and low-cost access to AI model APIs. The platform supports leading models including GPT, Claude, Grok, and Gemini, with GPT Pro rate multipliers as low as 0.1x. Join the group through the official website to receive $1 in trial credit.
APIMart	Thanks to APIMart for sponsoring this project! APIMart is a low-cost API platform for AI image & video generation — GPT-Image-2 from $0.006/image, 160+ images per dollar. One async API covers both image and video: submit a task, get an ID, fetch results via polling or callback. Batch tens of thousands of images without timeouts, switch models without changing code. Pay-as-you-go with no monthly fee — sign up here to get started.
UCloud ·星图AstraFlow	UCloud ·星图AstraFlow UCloud's AstraFlow large-model platform gives you one-click access to 200+ models: leading open-source models such as Kimi K3, DeepSeek V4/V3, Qwen 3, GLM5.2, and happyhorse are built in — no training of your own required, ready to use out of the box. Sign up with your email through the link above, complete real-name verification, and claim ¥50 in compute credits.
Fluxion AI	Thanks to Fluxion AI for sponsoring this project! Fluxion AI provides fast, reliable, and cost-efficient API access to GPT, Claude, Gemini, and other leading AI models through one unified API. New users can receive $3 in API credits through our dedicated link.
BeeAPI	Thanks to BeeAPI for sponsoring this project! BeeAPI is a professional multi-model AI API relay and aggregation platform that brings together many providers and a wide range of mainstream AI models. It supports cross-vendor price comparison, smart routing across groups, and unified API access and billing, helping you call AI services more flexibly, reliably, and at a lower cost.

Want to become a Codeg sponsor? Reach out to us by email.

✨ Features

One Interface for Every Agent Fifteen agents come built in, and any other ACP agent can join from the public registry or its distribution JSON. Codeg speaks the Agent Client Protocol to every one of them, so they all get the same rich conversation — tool cards, live diffs, plans, and permission prompts — instead of a terminal to squint at. Docs → · Custom agents →
Every Session, Searchable Codeg reads the history every agent CLI keeps on disk: import it in one click, search all of it, and resume any session where it stopped. @-mention an old session and the agent you're talking to can read it — even one a different agent wrote. Docs →
Multi-Agent Collaboration Type @, pick an agent, send. Each mentioned agent runs as its own session, side by side, streaming back into your thread — Claude Code drafting while Codex reviews — and sub-agents show up as cards that fill in while they work. Docs →
To-dos That Run Unattended Write a job down and walk away. Each to-do gets its own git worktree and branch, starts now or on a schedule, and waits in Review; accept the diff and the agent lands it, and Codeg checks git before calling it done. Docs →
Infinite Canvas Lay your work out in space: conversations, files, terminals, and notes become cards on a board, grouped by folder and agent, so several agents can run side by side in plain sight. Docs →
Desktop, Server, Mobile & Chat Run Codeg as a desktop app, a self-hosted server you open in any browser, or a single Docker container. Native iPhone, iPad, and Android clients — plus Telegram, Lark, and WeChat channels — keep your agents in reach, while your files and sessions stay on your own machine. Docs → · Mobile apps → · Chat channels →
Office Documents & Research Ask for a deck, a report, or a workbook and watch the real .pptx, .docx, or .xlsx render live beside the conversation. Bundled research skills — hypotheses, experimental design, statistics, literature search — work with any agent. Docs → · Research →
Browser & Computer Use In the desktop app, links open in a built-in browser beside your files: hand the agent a page, an element, a screenshot, or a console error, or let it drive a page you share. In preview, agents can also see and operate the windows you share — and Stop is one click away. Docs → · Computer Use →
The Full Engineering Loop An editor, live and side-by-side diffs, a full git client with a three-pane merge editor, one-click worktrees, an embedded terminal, and split views — all next to the agent, on the real files in your repo. Docs → · Git & worktrees →

Also in the box:

• Repository Panel — issues, pull requests, checks, and merges for GitHub, GitLab, Gitea, and Forgejo; any issue or PR can become a to-do
• Automations — save a configured composer and run it headlessly, on a cron schedule or on demand
• Token Usage — trends, cache hit rate, an activity heatmap, and usage by folder, agent, model, and session
• Fork & Steer — fork a conversation from a finished reply, or send a message into a turn that is still running, on agents that support it
• Skills & MCP — skill packs per agent, a local MCP scan, and registry search and install
• Project Boot — scaffold a new project visually, with a live preview
• Make It Yours — twelve themes you can recolor token by token, wallpapers, corner radius, and custom CSS — in any of ten interface languages
• Backup & Sync — encrypted backups, and configuration sync between machines through a file or your own WebDAV server
• URL Scheme — codeg://session/<id> opens a conversation from another app (desktop)
• And more — nearly every release adds something; the release notes are the full list

🤖 Supported Agents

Codeg talks to every agent over the Agent Client Protocol, so each one gets the same structured interface. Fifteen come built in, and Codeg installs, pins, and updates most of them for you:

Claude Code   Codex   Gemini CLI   OpenClaw   OpenCode   Cline   Hermes Agent   CodeBuddy   Kimi Code   Pi   Grok   Cursor   DeepSeek Harness   Qoder   Google Antigravity   + any ACP agent

Not on the list? Pick any agent from the public ACP registry or paste its distribution JSON — Codeg installs it, checks that it launches, and treats it like a built-in. Runtime requirements and where each agent keeps its sessions: Supported Agents.

📦 Install

Desktop — installers for macOS, Windows, and Linux are on Releases; setup steps in Installation.

Server — run Codeg headless and reach it from any browser. On Linux or macOS:

curl -fsSL https://raw.githubusercontent.com/xintaofei/codeg/main/install.sh | bash
CODEG_STATIC_DIR=/usr/local/share/codeg/web codeg-server

On Windows, in PowerShell:

irm https://raw.githubusercontent.com/xintaofei/codeg/main/install.ps1 | iex
$env:CODEG_STATIC_DIR="$env:LOCALAPPDATA\codeg-server\web"; codeg-server

Docker — the same server, in one container:

docker run -d -p 3080:3080 -v codeg-data:/data ghcr.io/xintaofei/codeg:latest

Mobile — install the iOS app or the Android APK, then point it at the Web Service of your desktop app or at your own codeg-server: URL, token, done. Pairing steps in Mobile apps.

Compose, prebuilt binaries, source builds, and in-place updates are covered in Deployment; environment variables in Configuration.

👥 Community & Support

• WeChat — scan the QR code to join our group for discussions, feedback, and updates:

  

• Issues — found a bug or missing a feature? Open an issue.

• Privacy — local-first: parsing, storage, and project operations stay on your machine, and web and server modes are guarded by token authentication. Details in Privacy & Security.

• LinuxDO — thanks to the LinuxDO community for their support.

• Show support — star the repo to follow along.

🤝 Contributing

Issues and pull requests are welcome. To build Codeg yourself, start with Development and Architecture.

⭐ Star History

🙏 Acknowledgments

• Agent Client Protocol — the foundation that lets Codeg connect to every agent it supports
• Superpowers — powers Codeg's expert skills module
• OfficeCLI — powers Codeg's Office documents workflow
• scientific-agent-skills — powers Codeg's Scientific Research skills (MIT-licensed subset)

📜 License

Apache-2.0. See LICENSE.