CodeConductor

January 26, 2026 · View on GitHub

CodeConductor banner

Version

A modern Desktop/Web UI for command-line AI agents
User-friendly | Visual interface | Multi-model support | Local-first data

Open-source enhanced fork of AionUI / Anthropic Cowork
_{CodeConductor builds upon the excellent AionUI foundation, adding workflow enhancements for real-world coding agent usage.}

English | 简体中文 | 日本語 | Official Website | Twitter

Positioning · What CodeConductor can do · Enhancements in this fork · Core features · Quick start · Documentation · Community and support

Changelog

1.7.16

Enhancement: Update Mem0 service to use ProcessConfig and v2 API for improved memory management.

1.7.15

Feature: Add Mem0 memory service integration for persistent agent memory.
Feature: Implement CLI installation management for streamlined tool setup.
Feature: Introduce image generation tool integration.
Fix: Update tag handling logic in build-and-release workflow.

1.7.13

Feature: Add custom assistant creation and management.
Feature: Add agent selection and categorization in the Guid page.
Feature: Add skill deletion support.
Fix: Initialize theme using DEFAULT_THEME instead of system detection.
Chore: Update About modal links for new repository ownership.
Chore: Update electron-builder config for improved installer behavior.

1.7.12

Added ActionToolbar to improve conversation interactions.
Added CLI model list hook and manual selection configuration handling.
Improved message processing and logging for easier debugging.

1.7.8

Enhanced chat UI components with improved visual design and message styling.
Introduced Superpowers Workflow service with guided agent setup page for workflow management.
Added GitHub Actions workflow for automatic Jira description synchronization on pull requests.
Improved collaboration UI with refined message display and interaction patterns.

1.7.7

Implemented Superpowers integration with comprehensive settings and command management.
Added dedicated settings page for configuring Superpowers modes and workflows.
Enhanced project workflow capabilities with Superpowers framework integration.

1.7.6

Fixed auto-updater 404 error by properly generating and uploading latest.yml metadata file.
Improved update mechanism reliability for automatic application updates.

1.7.5

Integrated popup-mcp as a built-in tool for interactive prompts and confirmations.
Added internationalization support with multiple language files.
Added CLI provider configuration UI and backend service for Claude and Codex.
Added storage utilities and a thought display component.
Fixed default value for ANTHROPIC_MODEL in CLI provider settings.
Fixed duplicate model setting calls in Guid component.

1.7.3

Enhance ACP adapter to better convert agent thoughts into displayable chat messages.
Implement CLI provider settings with presets and storage for Codex and Claude.
Improve collaboration features with multi-role communication and notification mechanism.
Optimize process management and orphaned process cleanup for better resource handling.
Enhance IPC communication mechanisms for improved stability.
Add electron-updater integration for automatic incremental updates.
Improve build and release pipeline with GitHub Actions automation.

1.7.2

Add a settings page for configuring CLI providers (Claude Code / Codex).
Add CollabChat view and MessageList with collapsible tool message batches for better readability.
Improve terminal and process handling (PTY-based terminal management, agent-browser command execution, worker process management).
Fix documentation links/addresses and clarify fork positioning; add Contributor Covenant Code of Conduct.

1.7.1

Add mention support in SendBox components (mention options, keyboard navigation, collab integration).
Improve UI/UX styling and theme consistency (inputs, buttons, modals, message visuals).
Enhance browser automation capabilities via agent-browser (IPC support and slash command entry).
Update build/docs/config templates and collaboration role guidelines.

This repository is an enhanced fork of CodeConductor. It keeps the original goal (a modern UI for command-line AI agents) and focuses on improving day-to-day “coding agent in real projects” workflows.

What CodeConductor can do

Multi-agent mode (unified UI for CLI tools)

If you have installed command-line tools like Claude Code, Codex, Qwen Code, Goose CLI, CodeConductor can detect them and provide a unified graphical interface.

Auto detection and unified UI - Recognize local CLI tools and bring them into one interface.
Local storage and multi-session - Save conversations locally and run multiple sessions with independent context.

Multi-Agent Mode Demo

Smart file management

Batch renaming, automatic organization, smart classification, file merging

Auto Organize: Intelligently identify content and auto-classify, keeping folders tidy.
Efficient Batch: One-click rename, merge files, say goodbye to tedious manual tasks.

Smart File Management Demo

Preview panel

Supports 9+ formats of visual preview (PDF, Word, Excel, PPT, code, Markdown, images, HTML, Diff, etc.)

View results instantly - After AI generates files, view preview immediately without switching apps
Integrated in Workspace - Preview is rendered inside the Workspace panel (right side), no separate middle panel
Real-time tracking and editing - Automatically tracks file changes; supports real-time editing of Markdown, code, and HTML

Preview Panel Demo

Image generation and editing

Intelligent image generation, editing, and recognition, powered by Gemini

AI Image Generation Demo

Multi-task parallel processing

Open multiple conversations, tasks don't get mixed up, independent memory, double efficiency

Conversation Management Demo

WebUI mode

Remotely control your AI tools - Access CodeConductor from any device on the network! Securely control local Claude Code, Codex, and other tools, data never leaves your device

# Basic startup
CodeConductor --webui

# Remote access (accessible from other devices on the local network)
CodeConductor --webui --remote

For full startup instructions on all platforms, see WEBUI_GUIDE.md.

Enhancements in this fork

This fork focuses on making command-line coding agents easier to use in real projects and daily work:

Multi-role collaboration view (PM/Analyst/Engineer) - Enable structured collaboration inside one project conversation.
Safer high-impact actions with explicit approvals - Clearer confirmation and visual cues for actions like executing commands or applying changes.
Slash commands for faster project workflows - Discoverable and reusable shortcuts for common actions in project chats.
Project-first conversations with workspace context - Task-driven, artifact-oriented project sessions with workspace context.
Review-friendly workflow - Preview and diff are treated as first-class citizens to reduce review friction.

Core features

Multi-session chat

Multi-Session + Independent Context - Open multiple chats simultaneously, each session has independent context memory, no confusion
Local Storage - All conversations are saved locally and will not be lost

Multi-model support

Multi-Platform Support - Supports mainstream models like Gemini, OpenAI, Claude, Qwen, flexible switching
Local Model Support - Supports local model deployment like Ollama, LM Studio, select Custom platform and set local API address (e.g., http://localhost:11434/v1) to connect
Gemini 3 Subscription Optimization - Automatically identifies subscribed users, recommends advanced models

File management

File Tree Browsing + Drag & Drop Upload - Browse files like folders, support drag and drop files or folders for one-click import
Smart Organization - You can let AI help organize folders, automatic classification

Preview panel

9+ Format Preview - Supports PDF, Word, Excel, PPT, code, Markdown, images, etc., view results immediately after AI generation
Real-time Tracking + Editable - Automatically tracks file changes, supports real-time editing and debugging of Markdown, code, HTML

Image generation and editing

Intelligent Image Generation - Supports multiple image generation models like Gemini 2.5 Flash Image Preview, Nano, Banana
Image Recognition & Editing - AI-driven image analysis and editing features

WebUI remote access

Cross-Device Access - Access from any device on the network via browser, supports mobile devices
Local Data Security - All data stored locally in SQLite database, suitable for server deployment

Personalized interface customization

Customize with your own CSS code, make your interface match your preferences

CSS Custom Interface Demo

Fully Customizable - Freely customize interface colors, styles, layout through CSS code, create your exclusive experience

Documentation

WebUI startup guide: WEBUI_GUIDE.md
Project overview and architecture notes: CLAUDE.md
Code style and conventions: CODE_STYLE.md

Quick start

System requirements

macOS: 10.15 or higher
Windows: Windows 10 or higher
Linux: Ubuntu 18.04+ / Debian 10+ / Fedora 32+
Memory: Recommended 4GB or more
Storage: At least 500MB available space

Download

Installation (desktop app)

Download and install the CodeConductor application
Configure your AI provider in Settings (Google account login or API Key, depending on provider)
Start a conversation and work in a project workspace

Run from source (developers)

npm install
npm start

WebUI mode (developers or headless)

npm run webui
npm run webui:remote

Community and support

Community

Your ideas and feedback are welcome. Use Discussions for ideas and issues for bugs or feature requests.

GitHub Discussions: https://github.com/zhu1090093659/CodeConductor/discussions
Issues: https://github.com/zhu1090093659/CodeConductor/issues
Releases: https://github.com/zhu1090093659/CodeConductor/releases

Contributing

Welcome to submit Issues and Pull Requests!

Fork this project
Create a feature branch (git checkout -b feature/AmazingFeature)
Commit your changes (git commit -m 'Add some AmazingFeature')
Push to the branch (git push origin feature/AmazingFeature)
Open a Pull Request