English | 简体中文

CodeNexus

Desktop workspace for Codex on Windows — built on the shoulders of giants.

CodeNexus brings Codex sessions, approvals, file changes, workspace context, and configuration into a focused desktop client.

Download Latest Release · Screenshots · Highlights · Development · Contributing

---

Overview

CodeNexus was built to make Codex app-server practical as a self-managed desktop workspace, rather than keeping the entire workflow inside a terminal. It focuses on the parts that matter during real agent work: understanding what happened, reviewing what changed, and keeping the local workspace under the user's control.

The app handles Codex app-server notifications directly and turns them into a desktop experience with timeline review, approval handling, custom themes, custom notification sounds, workspace file browsing, and a drag-friendly editor for local files.

CodeNexus also supports extending the agent runtime by injecting dynamic tools. The built-in codenexus.image_generate tool is wired into that path, so image generation can appear as a native part of the conversation and timeline instead of a separate external workflow.

CodeNexus is not an official OpenAI product. It is an independent desktop interface for Codex-oriented workflows.

Screenshots

Project screenshots are stored under docs/screenshots/.

Chat timeline

Workspace and file changes

Image generation workspace

Settings

Recent Updates

CodeNexus now visualizes streaming output from the Codex protocol, including command/process output deltas and streaming file-change updates. This makes long-running tool calls and patch activity easier to follow while a turn is still in progress.

This capability depends on Codex experimental protocol events. Enable the streaming output experimental feature in Settings before using it.

Highlights

Area	What CodeNexus Provides
Sessions	Start and continue Codex threads in a persistent desktop workspace.
Timeline	Review protocol events, command activity, approvals, diffs, MCP calls, and system messages in context.
Workspace	Browse project files, open multiple editor tabs, save changes, and inspect agent edits visually.
Approvals	Handle command, patch, and permission requests through desktop-native review surfaces.
Settings	Manage providers, models, skills, MCP, notifications, theme, fonts, and update behavior from one place.
Windows Experience	Installer-oriented packaging, desktop window lifecycle handling, local path support, and update integration.

Requirements

Dependency	Requirement
Operating system	Windows 10 or Windows 11
Node.js	Current LTS recommended
Package manager	pnpm@10
Codex CLI baseline	@openai/codex@0.131.0
Configuration	CC Switch recommended for Codex CLI

For Codex CLI provider, model, account, and environment configuration, use CC Switch, an all-in-one desktop manager for Claude Code, Codex, Gemini CLI, OpenCode, OpenClaw, and related agent tools.

After configuration, make sure Codex CLI is available:

codex --version

Development

Install dependencies and start the desktop app in development mode:

pnpm install
pnpm run dev

Run local verification:

pnpm run format:check
pnpm run lint
pnpm run typecheck

Contributing

Contributions are welcome through pull requests. For code changes, please keep the scope focused, describe the user-facing behavior, and run the local checks before opening a PR.

Release publishing is handled by project maintainers through GitHub Actions. Contributors should not create release tags for normal PR work.

For local community discussion, a QQ group QR code can be added to the Chinese README when the image asset is available.

Project Structure

Path	Purpose
src/main	Electron main process, windows, IPC, and service orchestration.
src/preload	Secure contextBridge preload layer.
src/renderer	Vue interface, Pinia stores, and runtime coordination.
src/shared	Cross-process contracts and protocol types.
scripts	Development, build, icon, and utility scripts.
music	Built-in notification audio resources.

Boundaries

• CodeNexus does not provide an OpenAI account, API token, hosted service, or model access.
• Model usage costs, workspace data handling, and local security remain the responsibility of the user.
• Third-party dependencies and bundled assets follow their respective upstream licenses.

License

MIT. See LICENSE.