Help Center/ CodeArts Agent/ User Guide (Extensions)/ Settings/ Basic Settings
Updated on 2026-07-30 GMT+08:00
View PDF
Share

Basic Settings

Click in the upper right corner of the CodeArts Agent chat panel to open the settings page.

The configuration items vary depending on the tool. This section uses Visual Studio Code as an example.

General Settings

Table 1 General settings

Item

Description

UI Language

Select the display language. If set to Auto, it will automatically adapt to the language settings of IntelliJ IDEA.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Font Size

Set the font size of the CodeArts Agent page.

Show Gutter Icons

Gutter icons are shortcut icons displayed next to definition lines of functions, classes, and other elements in the gutter area of the code editor. They are used to quickly perform common development operations.

Select the desired features (such as code explanation, commenting, optimization, and unit testing) from the drop-down list. The selected features will be displayed in the gutter area next to the corresponding code line. Click the icon and select the corresponding function to quickly perform operations on functions, classes, and other elements, improving development efficiency.

Minimum Log Level

Choose the minimum log level needed to maintain both system efficiency and effective troubleshooting.

  • Configuration principles: You are advised to set the log level to Debug. Debug logs record internal code execution details during development and debugging, making it easier to locate and troubleshoot problems that occur during code execution.
  • Log level: Log levels increase in severity like this: Debug, Info, Warn, and Error.
    • Debug: Records detailed internal information during runtime, such as request details and process steps. This level is mainly used for debugging and fault locating in the development phase.
    • Info: Records key status information during normal runtime, such as service startup and request processing summaries, to help you understand the basic system workflow.
    • Warn: Records potential issues that may affect system stability, such as missing configurations and insufficient resources. These issues do not directly cause program interruption.
    • Error: Records errors that have caused some functions to be abnormal or operations to fail, such as API calling failures and data parsing errors. These errors require timely attention and rectification.

Code Completion

Table 2 Code completion settings

Item

Description

Code Completion

When enabled, CodeArts Agent provides code completion features, such as real-time suggestions while typing and full code generation via keyboard shortcuts.

Native Completion Conflict Management

When enabled, the IDE's native completion may obscure the suggestions from this product. You are advised to keep this function disabled.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Do Not Disturb Mode

When enabled, code completion and chat-related prompts will be hidden to help you focus on coding. You can select specific options based on your needs.

In CodeArts Agent IDE and Visual Studio Code, the corresponding configuration items are as follows:

Hide Code Generation Shortcut Hints/Hide "Shortcut Key Prompt" Tips: When enabled, shortcut hints will no longer appear when code is selected.

In JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion), the corresponding configuration items are as follows:

  • Hide "Inline Edit" and "Add to Chat" Prompts: When enabled, the "Inline Chat" and "Add to Chat" prompts will no longer appear when code is selected.
  • Hide Inline Shortcut Hints: When enabled, inline shortcut hints related to code completion will not be displayed.
  • Hide Completion Toolbar: When enabled, the action toolbar displayed after code completion is generated will be hidden.

Inline Generation

When enabled, CodeArts Agent instantly suggests completions for the current line as you type code (before hitting Enter). These suggestions appear within 75 to 1,000 ms.

You can also adjust the debounce delay to prevent accidental triggers. The debounce delay (ms) sets the waiting time before triggering inline generation (ranging from 75 to 1,000 ms).

NOTE:

The tool stops suggesting code if the system faces heavy loads or encounters complex code contexts.

Snippet Generation

When enabled, CodeArts Agent will suggest ways to complete the smallest code block as soon as you hit Enter.

In JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion), you can also adjust the debounce delay to prevent accidental triggers. The debounce delay (ms) sets the waiting time before triggering snippet generation (ranging from 10 to 1,000 ms).

NOTE:

The tool stops suggesting code snippets if the system faces heavy loads or encounters complex code contexts.

Intelligent Truncation

When enabled, CodeArts Agent detects the code context and creates full code snippets automatically.

Java, C, C++, and Python are currently supported.

Continuation Context Cache

When enabled, cross-file context is automatically cached to avoid repeated extraction and speed up continuation response.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Format After Accept

When enabled, the accepted code will be formatted according to the current IDE code style to maintain consistency.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Dialog Flow

Table 3 Dialog flow configuration items

Category

Item

Description

Chat Settings

Response Language

This setting determines the response language for the model during chats. Choose a specific language or let the model adapt to the user's input language automatically, ensuring flexibility across various situations. Available choices include:

  • Auto: The model detects the question's language and responds accordingly.
  • Chinese: The model responds exclusively in Chinese, no matter what language the question uses.
  • English: The model responds exclusively in English, no matter what language the question uses.

Send Shortcut

Press Enter to send a message, Ctrl+Enter to start a new line: After entering a prompt in the input box, press Enter to send it. If you want to wrap the line and continue typing, press Ctrl+Enter.

Press Ctrl+Enter to send a message, Enter to start a new line: After entering a prompt in the input box, press Ctrl+Enter to send it. If you want to wrap the line and continue typing, press Enter.

Voice Settings

Customize the language displayed after speech-to-text conversion. Select one of the following options from the drop-down list:

  • Chinese: Regardless of whether the voice input is in Chinese, English, or a mix of both, the system converts it into the corresponding language (Chinese, English, or mixed of both).
  • English: Regardless of whether the voice input is in Chinese, English, or a mix of both, the system automatically converts it into English.
NOTE:

This configuration item is displayed only in Visual Studio Code.

Agent

Terminal Command Running Mode

You can select a policy for the agent to execute terminal commands based on your security requirements. By default, all terminal commands must be manually approved by the user before execution.

  • Running in Sandbox: Operations are performed automatically in a secure sandbox by default.
  • Auto Running: The agent directly executes all commands without your approval.

    For security purposes, you are advised to select Auto Running only when necessary. In this mode, the agent bypasses all security checks and may perform high-risk operations without notification.

  • Manual Running: Before executing any command, the agent sends a confirmation prompt to you. You need to manually confirm it.

Terminal Command Running Mode > Command Whitelist

This allows you to add prefixes of specific commands to the whitelist as required. Commands added to the whitelist bypass the sandbox mechanism and are executed outside the sandbox.

NOTE:

This option is displayed only when the running mode is set to Running in Sandbox.

Terminal Command Running Mode > Custom Sandbox Settings

By configuring the sandbox.json file, you can customize files and the network access scopes for processes within the current project's sandbox environment.

NOTE:

This configuration item is displayed only when Terminal Command Running Mode is set to Running in Sandbox.

Terminal Command Running Mode > Network Access

The agent provides three network access policies for you to choose from. Configure access control rules based on actual service requirements and security levels.

  • Network-wide: Access to all internal and external network resources is allowed.
  • Local: Only the local network (LAN/intranet) can be accessed. Access to the Internet is not allowed.
  • Blocked: All network connections are blocked, and access to internal or external resources is prohibited.
NOTE:

This option is displayed only when the running mode is set to Running in Sandbox.

Auto-approve > Edit

The agent can call tools like edit, write, and deleteFile to edit files on your computer.

Auto-approve > Open Browser

The agent can access websites in a browser.

Auto-approve > Web Crawler

The agent can access and capture specified web page content.

Agent Security Policies

(This option is displayed only when the running mode is set to Auto Running or Manual Running.)

Dotfile Protection (Dotfiles)

When enabled, the agent is not allowed to modify configuration files starting with a dot, such as .env and .gitignore.

Edit Range Protection

When enabled, the agent can only add, modify, and delete files in the current project directory. It cannot edit files in other directories.

Read Range Protection

When enabled, the agent can only access and read files in the current project directory. It cannot access files in other directories.

Ask

Suggested Questions

When enabled, the system predicts three relevant questions using your preferences and chat history after answering your question. Click these suggestions to ask them instantly.

Agent

Agents are programming assistants tailored for diverse development scenarios. CodeArts Agent not only provides various built-in agents but also supports custom agent capabilities. By configuring prompts, skills, and tool sets within a custom agent as required, you can create a dedicated assistant to handle complex development tasks more efficiently.

For details about built-in agents, see Built-in Agents. For details about custom agents, see Custom Agents.

Models

In addition to built-in models, enterprise members can also connect third-party large language models to CodeArts Agent. If you have purchased the CodeArts Agent professional edition, you can also add third-party models on the console.

For details, see Models.

MCP Tool

CodeArts Agent connects to an MCP server using the Model Context Protocol (MCP). It leverages the server's extra tools and resources to enhance its capabilities. You can add an MCP server. For details, see MCP.

Skills and Rules

You can view and configure skills and rules for enterprises, teams, projects, and users. This ensures consistent yet adaptable workflows tailored to general standards or individual preferences. In addition, multiple out-of-the-box skills are available in the marketplace to quickly boost development efficiency.

  • Enterprise skills and rules: Standardized skill libraries and business rules that are centrally built, managed, and distributed at the company or large organization level. The core objective is to eliminate execution deviations and ensure that different teams and individuals strictly adhere to unified operation standards, security protocols, and process guidelines in complex business scenarios. This enhances overall operational consistency and controllability.
  • Team skills and rules: Skills and rules that are uniformly configured for teams or departments to standardize team operations and ensure consistent operations among all members. Only team administrators or department administrators can create, modify, and delete these skills and rules. Common members can only use them.
  • Project skills and rules: Skills and rules that apply only to the current project. Bound to the project repository, they are distributed alongside it to address specific project needs.
  • User skills and rules: Skills and rules configured based on individual habits or preferences. These settings take effect only for the user who configures them.

For details about skill configuration, see Skills. For details about rule configuration, see Rules.

Memory

Memory allows CodeArts Agent to adapt to how you work. The agent will record your habits and preferences, such as your preferred answer style and language.

For details, see Memory.

Repository Index

CodeArts Agent supports repository indexing, helping you quickly search for and obtain related information from the repository. Repository indexes are classified into personal indexes created locally and team indexes created on the cloud.

For details, see Repository Index.

Lab Features

Table 4 Lab feature settings

Item

Description

Code Optimization

When enabled, the agent automatically identifies issues such as duplication, complexity, performance, and security in the project, and completes structured optimization and repair in one click.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Test Case Repair

When enabled, compilation and execution issues in test cases can be fixed.

NOTE:

This configuration item is displayed only in JetBrains tools (PyCharm, IntelliJ IDEA, WebStorm, and CLion).

Local LSP

When enabled, the agent can leverage diagnostics, definitions, and symbol data from language services for more precise code comprehension and modification.

Language Server Protocol (LSP) is a standardized protocol for providing language services. By decoupling language services from the IDE, LSP enables an independent language server process to deliver core capabilities such as code analysis, error diagnosis, fast navigation, and code refactoring.

For details about LSP, see LSP.

Knowledge Base

CodeArts Agent enhances chat capabilities by utilizing knowledge bases as context. Developers can upload internal documents and technical resources to improve response accuracy.

For details, see Knowledge Bases.

Parent Topic: Settings