Run Claude Code Inside IDEA Without Switching Terminals – CC GUI Adds Codex Too

Why CC GUI?

Developers using JetBrains IDEs often face a fragmented workflow: AI agents work well but require terminal switches, disrupting coding rhythm. CC GUI (Claude Code GUI) solves this by placing Claude Code and Codex side‑by‑side in the IDE sidebar, enabling conversation, command execution, session handling, and vendor configuration without leaving the editor.

Emerging Trend

AI‑assisted programming is shifting from simple code‑completion plugins to full‑cycle IDE agents that can plan, modify, verify, roll back, and iterate on real projects. The value of tools like CC GUI lies in keeping the entire agent workflow inside the IDE, reducing context‑switching overhead.

Core Capabilities

Dual‑Engine Support : Switch between Claude Code and OpenAI Codex within the same plugin.

Intelligent Dialogue : Supports @‑file references, image sending, conversation rollback, and prompt enhancement.

Agent Commands : Built‑in slash commands such as /init and /review, extensible via MCP.

Development Experience Optimizations : Code diff, permission management, file navigation, theme switching, and automatic Chinese/English handling.

Session Management : History search, bookmarking, export, vendor switching, and usage statistics.

Getting Started Inside IDEA

1) Install the Plugin

Search for “Claude” in the IDEA Plugin Marketplace and install CC GUI.

2) Add SDK Dependencies

After the first launch, open Settings → SDK Dependencies and download the Claude Code SDK or the Codex SDK. Versions can be updated later from the same location.

3) Configure a Provider

In the panel’s top‑right “Provider Management”, add the following fields:

Name

API Key

Endpoint URL

Model ID mapping

After saving, the provider is ready. The article’s example uses a “Coding Plan” from the service Code80.

4) Run a Task

Select Claude Code or Codex in the chat window and start with the command /init. The agent scans the project, generates a CLAUDE.md file, and establishes a unified context for subsequent actions. The author demonstrates upgrading a project from Vite 7 to Vite 8 by first producing a Markdown plan and then applying the changes.

Practical Value for Teams

Reduced Switching Cost : No need to jump between IDE and terminal.

Lowered Entry Barrier : Friendly for developers unfamiliar with command‑line tools.

Preserved Deep Agent Capabilities : Supports project‑level task flows rather than a simple chat panel.

When developers are already accustomed to JetBrains’ engineering experience, integrating agents in‑place feels smoother than rebuilding the toolchain.

What Is Claude Code?

Claude Code is positioned as an executable programming agent rather than a mere autocomplete tool. Its abilities include:

Reading and writing files

Executing shell commands

Cross‑file modifications

Running tests and iteratively fixing issues

Official subscriptions are offered as a Pro tier (monthly) and a higher tier starting at $100/month, but they require overseas credit cards and stable network access.

Alternative Access

For users in regions with payment or connectivity constraints, the service Code80 provides a shared subscription account that can be used via a custom endpoint, delivering the same experience as the official API.

FAQ

Q1: Is CC GUI limited to Claude?

A: No. It supports both Claude Code and Codex, allowing engine switching per task.

Q2: Why can’t I use the plugin immediately after installation?

A: The first use requires downloading the appropriate SDK (Claude Code SDK or Codex SDK) and configuring provider parameters.

Q3: What does the /init command do?

A: It analyzes the project structure, generates a CLAUDE.md file, and creates a unified context for later commands.

Q4: Who should use this plugin?

A: Developers who spend most of their time in the IDEA/JetBrains ecosystem and want to avoid terminal switches while retaining deep agent capabilities.

Q5: How to access Claude Code more smoothly in China?

A: Use Code80 to obtain a ready‑made subscription and endpoint, bypassing credit‑card and network hurdles.

Engineering‑Centric AI Workflow (Vibe Coding)

The author’s Vibe Coding approach treats AI programming as an engineering process, using custom commands as the workflow skeleton: /commit – standardizes commit flow /upstream – synchronizes branches in minutes /progress-save / /progress-load – prevents context loss /deploy – one‑click deployment /gitsync – reliable multi‑project code sync /review and /bug-add – ensure quality and knowledge capture /parallel-epic – enables parallel multi‑agent development

These markdown‑based commands are simple to write, yet combined they let developers focus on “what to do” while the agent handles “how to do it”.