Qoder CN Series:Qoder CN Update Log

Feature updates

• Introduced Qoder CN Exclusive Edition. You can log on to Qoder CN Desktop, Qoder CN CLI, and QoderWork CN with a single account.

• The Experts team now supports Qwen 3.7-Max and GLM-5.2.

• The Computer Use feature now supports Windows. The Computer Use feature is now available on Windows devices.

• RepoWiki generation is now more controllable. RepoWiki generation now supports pre-intervention, which lets you specify the analysis scope and focus before generation. You can use interactive dialogs to create a plan file that guides the RepoWiki creation process, which enables human-machine collaboration for generating, modifying, and rewriting knowledge.

• Quest now supports sessions without a workspace. You can start a session directly without opening a workspace, which is ideal for quick questions and lightweight tasks.

• Quest tasks support quick navigation to conversations. The Quest window now includes a side navigation feature that lets you quickly navigate to specific conversation snippets.

Experience optimizations

• Optimized the logon prompt for Qoder CN IDE to automatically remember your last logon method.

Experience optimizations

• Optimized the logon guide for Qoder CN IDE.

Experience optimizations

• Optimized the model selector by adding model descriptions.

• Optimized the page prompt for upgrading a personal subscription to the Qoder CN suite.

The Qoder CN product suite now uses a unified Credits system for personal users across all products. You can use the same Credits balance for AI-assisted coding in Qoder CN IDE and intelligent desktop collaboration in QoderWork CN. You no longer need to subscribe to and manage different products separately.

Feature updates

• The Quest window now supports custom layouts and a "My Quests" task board. You can drag and split panes in the Quest window to handle multiple tasks in parallel. The new "My Quests" board summarizes all Quests by status for a clear overview of your progress.

• The Experts team now supports custom models. You can flexibly configure the models used by the Experts team and sub-agents. You can switch between built-in models or use your own custom models to adapt to more task scenarios.

• The Computer Use and window snapshot features are now available on macOS. After you grant authorization, the agent can call computer operations on macOS and control apps in the background to complete complex automation tasks. You can double-press the ⌘ Command key to capture the frontmost window as context for the agent, which enables more accurate understanding. This feature is for macOS only.

• Dev Container is now supported in SSH Remote. Dev Container support is added for SSH Remote scenarios. You can also attach to any Docker container, which improves the flexibility and consistency of your remote development environment.

• Enhanced code review capabilities. Support for `/ultra review` is added to improve the depth of analysis and the quality of suggestions in code reviews. This helps you find potential issues and improve implementations faster.

Fixes

• Fixed known issues.

Feature updates

● Added `/canvas` for visualization generation. You can now use `/canvas` to generate interactive Canvas views, which makes content organization, report presentation, and analysis more intuitive.

● Enhanced code review capabilities. Support for `/ultra-review` is added to improve the depth of analysis and the quality of suggestions in code reviews. This helps you find potential issues and improve implementations faster.

● Added resource plans for the Enterprise Standard subscription of the Credits-based billing system. Enterprise administrators can now set usage limits on resource plans for members.

Fixes

• Fixed known issues.

Fixes

• Fixed known issues.

Fixes

• Fixed an issue with Credits usage display and refresh.

Feature updates

• Quest independent window and completely new Editor. Quest has been upgraded from an IDE internal schema to an independent window that serves as a command center for the Agent First paradigm. It supports running agent tasks for different projects in multiple workspaces simultaneously. A unified panel provides real-time tracking of global dynamics, and a delivery manifest is automatically generated upon task completion. The Editor window has been fully upgraded for a smoother human-computer collaborative coding experience.

• Experts team mode is now officially available. Multiple experts work in parallel for high-quality delivery, covering the entire process from solution design and coding to quality inspection. It supports an end-to-end pipeline from solution design to code delivery. The self-evolving AI team learns your coding style and accumulates exclusive collaboration experience over time. Five types of experts, including planning, research, coding, review, and testing, are now part of Quest to complete end-to-end delivery in a pipeline. You can also customize experts by configuring domain knowledge, skills, and tools to create a dedicated agent team that is tailored to your business scenarios.

• Added support for models such as GLM-5.1 and Kimi-K2.6. This provides more model choices for AI chat and agent modes.

Billing and entitlement changes

• Qoder CN (formerly Lingma) has been upgraded to a Credits-based billing system. All subscription plans deduct Credits based on actual usage. For more information, see Billing overview.

• Existing Enterprise Edition subscriptions will continue under the original price and entitlements for the current contract period. New seats or renewals are subject to the new pricing.

Feature updates

• You can now upgrade Lingma IDE to Qoder CN IDE.

  • Upgrade to Qoder CN to unlock a new experience.

    Quest independent window: Immersive conversations for a more focused workflow.

    Completely new Editor: A smoother human-computer collaborative coding experience.

    For Credits subscription users, new features include the Experts team and models such as GLM-5.1 and Kimi-K2.6.

  • Seamless and worry-free migration. Your historical sessions, rules, memories, agents, and code indexes are automatically retained.

Fixes

• Fixed known bugs.

Feature updates

• Code Review agent. The built-in Code Review agent can be triggered using `/code-review` within a session in agent mode to review a specified scope of code.

• Browser agent. The built-in Browser agent can access web pages and read online documents using the `/browser` command in agent mode.

• Hook mechanism. A new Hook mechanism has been introduced into the agent execution flow. It lets you insert custom logic at key nodes of agent execution and supports events such as UserPromptSubmit, PreToolUse, PostToolUse, PostToolUseFailure, and Stop.

Experience optimizations

• Clearer feedback for conversation rollback operations. Optimized the interactive feedback after a revert is triggered to help users more intuitively understand the rollback scope and current status.

Fixes

• Fixed known issues.

Feature updates

• DevContainer support. Added development container (DevContainer) support. You can define a standardized development environment through a `.devcontainer/devcontainer.json` configuration file to help teams achieve a consistent development experience.

• Agent mode now supports a terminal sandbox on Windows. Commands that are executed by the agent now run in a sandboxed container, which effectively isolates potentially risky operations.

Experience optimizations

• Extension Host process isolation. Optimized the process isolation for built-in AI coding-related extensions. AI extensions now run in a separate Extension Host process, which improves runtime stability and the overall user experience.

Feature updates

• Agent sessions can now be opened in a separate window. You can now open agent sessions in new windows for more efficient task handling in parallel multi-session scenarios, which further improves the user experience.

• Quest mode now supports editing sent messages. In Quest mode, you can now edit messages that have already been sent. This lets you adjust conversation content at any time for greater interaction flexibility.

• Lingma IDE now supports AccessKey/SecretKey (AK/SK) authentication. You can now log on using an AccessKey pair (AccessKey ID and AccessKey secret).

• The Spec/plan preview panel now includes a file search function that lets you quickly locate object files.

Feature updates

• Skill UI dynamic interface generation is now available. When an agent executes a Skill, it can render an interactive HTML component in real time within the conversation. This supports rich interactive scenarios such as forms, charts, and configuration panels. The first time you use this feature, you must have the agent create an interface for the corresponding Skill.

Experience optimizations

• Upgraded navigation in the Quest mode output area. Optimized the navigation bar and file tree hierarchy in the output area for a better interactive experience.

• Optimized the experience in Quest mode when no project is open to make it easier to find historical tasks.

• Optimized error code prompts for RepoWiki.

Fixes

• Fixed bugs related to issue reporting.

Feature updates

• Files in the Quest mode output area now support fine-grained modification with "Add to chat". You can select a portion of the content and add it to the conversation. This works for both code files and Spec files.

Experience optimizations

• Optimized the task list sorting in Quest mode to display tasks in descending order of the most recent conversation.

• Optimized network connectivity for the IDE plugin marketplace in the Lingma exclusive enterprise edition.

Experience optimizations

• Added an entry point in the model selector for configuring third-party models with Bring Your Own Key (BYOK).

Feature updates

• Support for custom models. Lingma Personal Professional Edition now supports connecting to the Coding Plan of providers such as Alibaba Cloud Model Studio, Zhipu, Kimi, and MiniMax through API keys to use third-party models.

• Agent mode now supports cross-project search and file editing within a Workspace.

• Upgraded the VS Code kernel to version 1.106.3.

• RepoWiki upgrade.

• Core experience upgrade. Upgraded the core experience with visualized generation progress, optimized operation guides to distinguish between generation and updates, added import/export instructions for two-way synchronization, and added an automatic export feature.

• More efficient traceability. Improved traceability by automatically displaying the Commit ID after generation for projects that are associated with Git. Added line numbers to source reference annotations.

• New user activity promotion. Displays the remaining number of free generations. Note: RepoWiki is currently in Beta for the Enterprise Edition.

Experience optimizations

• The input box now lets you add up to 20 images and 20 files.

• The file tree in Quest mode now supports right-click operations, such as Add to context, Copy relative path, and Open in folder.

• Optimized the conversation interaction in agent mode for empty projects.

Feature updates

• Visual management for skills and custom agents. Provides a visual management interface for skills, custom agents, and instructions. This interface supports management operations such as automatic creation and import.

Experience optimizations

• Expanded the configuration items for custom agents to support skill configuration.

• Optimized the experience of opening code files in the Quest mode output area.

Feature updates

• Added built-in `create-skill` and `create-agent` skills to guide and simplify the creation of skills and custom agents. You can also use `/` to quickly invoke and use skills and sub-agents.

• Quest mode now includes a file tree to improve file browsing within the current project, which lets you easily add files to the context.

• Quest mode now supports invoking the Vercel Skill through conversation for quick application deployment.

Experience optimizations

• One-click prompt optimization. Improved handling of scenarios where the prompt contains `@context`.

• Support for the `.agents/skills` directory. You can now read skills from the `.agents/skills` directory, which adheres to industry standards.

• You can now view execution details for custom sub-agents in an expanded pop-up window.

• NEXT now supports reading clipboard history.

• Enhanced the ability of agents to call Skills in Quest tasks.

• Optimized the experience of creating new tasks and collapsing the toolbar on the Quest mode page.

• Updated the content of the use cases on the Quest mode home page.

• Logon stability optimization. Improved logon stability. If you encounter logon, permission, or other issues after you upgrade, you can try logging out and logging back in.

Fixes

• Fixed an occasional system exception issue.

• Fixed an issue where editing files would occasionally get stuck.

• Fixed an issue where including an image in the context could cause the IDE to crash.

Experience optimizations

• Optimized the processing logic in AI chat (Ask) mode when tools are not enabled.

Feature updates

• Major model upgrade to Qwen-Coder-Qoder. A deeply customized model that is designed to enhance the end-to-end programming experience. This model is based on Qwen-Coder and has undergone extensive reinforcement learning for the agent framework, tools, and scenarios. In our real-world software engineering task evaluations, it has shown excellent task resolution rates, especially with a 50% higher accuracy in terminal commands on Windows systems.

• Lingma IDE now supports Quest autonomous programming, which is available in Beta.

• Autonomous programming agent. The agent independently completes high-quality, end-to-end, deliverable results without continuous manual intervention.

• Requirement alignment mechanism. Clarifies the task before execution through intent recognition, requirement clarification, and co-creation of a Spec.

• Long-running task capability. Significantly improved the ability to sustain long-running tasks. The agent supervises the task until completion.

• Autonomous quality assurance. Integrated result validation capabilities to independently verify and fix the quality of deliverables.

• Continuous self-evolution. The agent not only executes tasks but also actively evolves. It remembers your style and continuously learns new technologies.

• Support for more application scenarios. Quest performs intent recognition on user input and automatically routes to the most suitable capability.

• Spec-driven development. Aligns requirements and constraints first, then executes and accepts tasks, which ensures clear and traceable results.

• From idea to product. Supports creating websites and prototypes from scratch and integrates top-tier visual models to significantly improve design quality and usability.

• Agentic Chat. Supports parallel execution by multiple agents, with a built-in planning agent and comprehensive upgrades to custom extension capabilities.

• New built-in planning agent. When you handle complex tasks, you can use a human-computer collaborative mode to create a feasible execution plan in advance. This can significantly improve task execution efficiency, avoid rework, and enhance the quality of agent-generated code and the controllability of results.

• Upgraded custom extension capabilities. Supports custom agents (sub-agents), skills, and commands to create your own intelligent workflow.

• Parallel multi-agent sessions. Supports running multiple session windows in parallel to handle different tasks simultaneously, which significantly improves multitasking efficiency.

Experience optimizations

• Optimized the logon prompt for Qoder CN IDE to automatically remember your last logon method.

Experience optimizations

• Optimized the logon guide for Qoder CN IDE.

New

• One account for the Qoder CN suite. You can use a single Qoder CN account to seamlessly log on to the IDE, JetBrains plugin, QoderWork CN (desktop), Qoder CN CLI (terminal), QoderWake CN (digital employee), Qoder CN Mobile, and Cloud Agents without separate registrations.

• Unified Credits pool. Your original account's Credits balance is converted to the new suite account at a 1:1 ratio. The validity period is reset to one month from the date of the upgrade.

• Limited-time model benefits. Upgraded users can enjoy 200 free calls to Qwen3.7-Max daily. This is a limited-time offer, and the end date will be announced on the official website.

• Complete AI data retention. Your historical session records, custom rules, memories, skills, agents, code indexes, and general plugin settings are all migrated to the new account without any data loss.

Changes

• Logon method switch. You can automatically upgrade using the "Log on with Alibaba Cloud account" option. After a successful upgrade, use the "Log on with Qoder CN account" option. The original logon method will no longer be available.

• VS Code plugin support discontinued. The original Qoder CN (formerly Lingma) VS Code plugin is now deprecated and is no longer part of the product suite. We recommend that you switch to Qoder CN IDE.

• Automatic order refunds. After the upgrade, fees for the next and subsequent billing cycles of your original subscription are automatically refunded through the original payment method. Auto-renewal orders are canceled at the same time. You do not need to apply for a refund.

Feature updates

• The AI chat now supports multiple tabs, which lets you quickly create, switch, and close sessions.

• Supports `.amd` file separator markers to help identify content that is used as input prompts.

• Added resource plans for the Enterprise Standard subscription of the Credits-based billing system. Enterprise administrators can now set usage limits on resource plans for members.

Fixes

• Optimized the product experience and fixed known bugs.

Feature updates

• Added support for GLM-5.1 and Kimi-K2.5 models. This provides more model choices for AI chat and agent modes.

Billing and entitlement changes

• Qoder CN (formerly Lingma) has been upgraded to a Credits-based billing system. All subscription plans deduct Credits based on actual usage. For more information, see Billing overview.

• Existing Enterprise Edition subscriptions will continue under the original price and entitlements for the current contract period. New seats or renewals are subject to the new pricing.

Feature updates

• Introduced the Qwen-Coder-Qoder model. A deeply customized model that is designed to enhance the end-to-end programming experience in Qoder. This model is based on Qwen-Coder and has undergone extensive reinforcement learning for the Qoder agent framework, tools, and scenarios. In our real-world software engineering task evaluations, its task resolution rate has surpassed that of Cursor Composer-1, with a 50% higher accuracy in terminal commands on Windows systems.

• Completion and Next Edit Suggestion (NES) have been fully upgraded to NEXT. Moving from passive completion to proactive prediction, it supports features such as automatic dependency import, with a comprehensive upgrade in both capabilities and interactive experience.

• You can now manually trigger NEXT code completion.

• Full support for `@database`. You can now reference databases as context in the Ask/Agent mode of Qoder.

• You can now generate and execute SQL in the Query Console.

• Supports compatibility with `AGENTS.md`. When an `AGENTS.md` file exists in the root directory of the current project, the agent automatically recognizes and loads the file content.

• Supports adding the `.aiignore.md` extension for restricted file support.

• Tools can now edit files outside the project.

Experience optimizations

• Optimized several product experience issues.

• Supports IntelliJ 2026.1 EAP+ versions.

Fixes

• Fixed several known issues.

Fixes

• Fixed some known bugs.

Fixes

• Fixed an issue of occasional abnormal CPU and memory consumption.

Feature updates

• Added the `@openFiles` annotation for context, which lets you add all open files at once, including project files, SDK files, and Scratch files.

• Added tools to enhance real-time awareness of file content in the edit area and optimize agent performance.

Fixes

• Fixed some known bugs.

Fixes

• Fixed bugs.

Fixes

• Fixed an issue where the logon status was lost.

Fixes

• Fixed an issue where the system paste behavior failed in CLion and Rider.

• Fixed an issue where the "Qoder CN One-Click Fix" feature occasionally failed.

• Fixed an issue where the default AI reply language occasionally switched to English.

Feature updates

• Supports enterprise model selection.

Fixes

• Fixed several bugs.

Fixes

• Fixed bugs.

Fixes

• Fixed bugs.

Fixes

• Fixed an issue where custom instructions failed to execute in specific scenarios.

• Fixed several other bugs.

Experience optimizations

• Optimized file editing access control for the exclusive edition.

Fixes

• Fixed bugs.

Feature updates

• Programming capability upgrade. Combined with the latest Qwen model capabilities, it supports parallel tool calling and deeply optimized coding performance, which makes complex coding tasks more efficient and accurate.

• New To-Dos tool. The programming agent can perform structured to-do breakdowns before it starts a coding task, which enables more effective execution of complex, long-running coding tasks.

• New message queuing. While Lingma is replying or executing a task, you can edit and send subsequent messages directly. This helps developers iterate quickly and continuously without waiting.

• New fully automatic execution mode. The programming agent supports configuring an automatic execution mode. When this mode is enabled, you can configure a blocklist for terminal commands, which significantly improves the efficiency of coding task execution while ensuring security.

Experience optimizations

• Enhanced project awareness. Optimized retrieval capabilities such as Grep and Search Codebase for more effective project searches.

• Enhanced memory awareness. Optimized memory and memory organization capabilities to make the responses from Lingma more personalized.

• Optimized tool presentation experience. Reduced the prominence of interface elements for tools such as file reading, web search, project search, and directory search. Strengthened confirmation-type operations to help developers obtain key information faster.

• The agent mode is now the default selection for AI chat.

• In AI chat, you can now add context by typing `@` to open a window.

• AI chat mode now supports search tool calling by default. You can disable this feature in the settings.

• You can now add selected code lines from the code editor directly to the conversation window as context using a right-click operation.

• You can now configure multiple AI rules for a project and their activation methods.

• Optimized the inline conversation editing experience to enhance the experience for file editing scenarios.

• Removed the model selection entry and file editing mode to simplify choices for developers.

Fixes

• Issue reporting now supports adding images.

Fixes

• Fixed bugs.

Experience optimizations

• Optimized the experience of the rollback link for inline code suggestions.

Fixes

• Fixed bugs.

Experience optimizations

• Optimized the generation effect of inline suggestion predictions.

• Optimized the information prompt for when file modification fails.

• Optimized the information prompt for path exception cases.

Fixes

• Fixed an issue in K2 mode where plugin installation occasionally failed and the function-level shortcut entry for Kotlin files was invalid.

• Fixed an issue where running Gradle clean/build on Android Studio caused the IDE to stutter.

• Fixed an issue where the input box occasionally became unresponsive.

Experience optimizations

• The MCP service now supports Streamable HTTP.

• Optimized the memory generation effect.

• Optimized some interface experiences in the MCP service.

Fixes

• Fixed bugs.

Feature updates

• Code snippets in AI chat responses now support one-click application to code files.

• Code issue fixing (Problem) now supports agent mode for repairs.

• The debug error fixing scenario is now optimized to be answered in agent mode.

Fixes

• Fixed several bugs.

Experience optimizations

• Optimized the MCP tool parameter settings experience.

• Optimized the focus prompt style of the input box area.

Fixes

• Fixed an occasional installation and decompression failure issue for Windows 7 users.

• Fixed several other bugs.

Experience optimizations

• The MCP marketplace now supports English descriptions.

Fixes

• Fixed an issue where AI chat mode could not follow language preferences after tools were enabled.

Feature updates

• AI chat (Ask) mode now supports tools such as project and code search, and web search.

• Agent mode now supports the web search tool.

• Agent mode now supports selecting a file directory (folder).

Experience optimizations

• Supports hybrid search for project files and folders when you add context, with search highlighting.

• Optimized the experience of pasting code snippets from the project into the AI chat window.

• Compatible with the latest IDEs' Reworked Terminal, which does not support prompts.

• Supports selecting instructions during a session.

• Optimized the loading speed of the knowledge base when you select teamDocs.

Feature updates

• New Next Edit Suggestion (NES) feature. Supports predicting the next modification suggestion. You can enable this feature in the plugin settings.

• MCP tool execution in the agent now supports configuring automatic execution.

Experience optimizations

• Optimized the experience of adding to the terminal command automatic execution whitelist. You can now add to the whitelist with one click during a session.

• You can now drag folders from the file list to the session window input box as context.

• Optimized file modification accuracy. Failed file modifications are automatically retried.

• Optimized the experience of pasting code from within the project into the session window input box.

• Optimized the code change generation effect during inline sessions.

Fixes

• Fixed an issue where inline sessions were not visible when they exceeded the editor area.

• Fixed an issue where the number of completions was missing from the plugin settings.

Experience optimizations

• Added a suggestion prompt for selecting a session mode.

• Supports one-click repair for MCP service connection exceptions.

• Optimized the output effect of the qwen3-thinking model in agent mode.

• Reduced the probability of flowchart generation under the code explanation instruction.

Fixes

• Fixed an issue where AI chat mode occasionally failed to perceive the currently open file.

• Fixed an occasional output compatibility issue when you switch between agent and file editing modes in project search.

• Fixed an occasional compatibility issue with PyCharm 2025.1 in file editing mode.

Feature updates

• New Inline Chat feature. Supports Q&A and code modification within the editor area.

Experience optimizations

• Optimized the effect of large file modifications in AI chat scenarios.

Feature updates

• Tool use. The agent can autonomously use more than ten built-in programming tools, such as reading and writing files, code querying, and error troubleshooting. It also supports MCP tool configuration and integrates with the ModelScope MCP marketplace, which allows developers to configure freely.

• Terminal command execution. When executing coding tasks, the agent can autonomously decide which commands to execute, and automatically write and run commands in the terminal, which significantly improves the efficiency of coding task execution.

• Project-level changes. Based on the developer's task description, the agent can autonomously break down tasks and modify multiple code files within the project. It also supports gradual iteration or snapshot rollbacks through multiple conversations, which allows for collaborative coding with Lingma.

• Automatic project awareness. Based on the developer's task description, the agent can automatically perceive project information such as the framework, technology stack, required code files, and error messages. This eliminates the need to manually add project context and makes task descriptions easier.

• Memory awareness. The agent supports autonomous memory capabilities that are based on a large language model. As developers converse with Lingma, it gradually forms rich memories related to the developer, project, and issues, and becomes more knowledgeable with use.

Experience optimizations

• You can now navigate to files, methods, and definitions within the project from the AI chat.

• You can now add project directories as context in AI chat.

• You can now specify a knowledge base when you add context in AI chat (Enterprise Edition).

• You can now multi-select files, directories, and gitCommits when you add context in AI chat.

Feature updates

• Added resource plans for the Enterprise Standard subscription of the Credits-based billing system. Enterprise administrators can now set usage limits on resource plans for members.

Fixes

• Fixed bugs.

Billing and entitlement changes

• Qoder CN (formerly Lingma) has been upgraded to a Credits-based billing system. All subscription plans deduct Credits based on actual usage. For more information, see Billing overview.

• Existing Enterprise Edition subscriptions will continue under the original price and entitlements for the current contract period. New seats or renewals are subject to the new pricing.

Fixes

• Fixed bugs.

Experience optimizations

• The Lingma VS Code plugin and JetBrains plugin versions are now isolated. In future versions, the logon status, session records, and data for the two plugins are kept separate.

Fixes

• Fixed bugs.

Fixes

• Fixed bugs.

Feature updates

• You can now configure multiple AI rules for a project.

Experience optimizations

• Optimized the header configuration experience for the MCP service.

• You can now click to view modified files in the session response.

• Optimized the speed of code completion scenarios.

• Optimized the information prompt for path exception cases.

Feature updates

• Optimized the search experience when you add context. You can now simultaneously search for files and file directories.

• You can now add open files as context.

Experience optimizations

• The MCP service now supports Streamable HTTP.

• Optimized the memory generation effect.

• Optimized some interface experiences in the MCP service.

Fixes

• Fixed bugs.

Feature updates

• The code issue fixing (Problem) scenario now supports being answered in agent mode.

Fixes

• Fixed several bugs.

Feature updates

• You can now drag files and folders to the session window input box as context by holding down the Shift key. This feature requires VS Code 1.100 or later.

Experience optimizations

• Optimized the MCP tool parameter settings experience.

• Optimized the experience of pasting code from within the project into the session window input box.

Fixes

• Fixed several bugs.

Experience optimizations

• The MCP marketplace now supports English descriptions.

Fixes

• Fixed an issue where AI chat mode could not follow language preferences after tools were enabled.

Feature updates

• AI chat (Ask) mode now supports tools such as project and code search, and web search.

• Agent mode now supports the web search tool.

• Agent mode now supports selecting a file directory (folder).

Experience optimizations

• Optimized the loading performance of the personal settings page.

Fixes

• Fixed an occasional account error issue in SSH Remote scenarios.

• Fixed an issue where pasted images were occasionally not recognized.

Feature updates

• MCP tool execution in the agent now supports configuring automatic execution.

Experience optimizations

• Optimized the experience of adding to the terminal command automatic execution whitelist. You can now add to the whitelist with one click during a session.

• Optimized file modification accuracy. Failed file modifications are automatically retried.

• Optimized the code change generation effect during inline sessions.

Experience optimizations

• Added a suggestion prompt for selecting a session mode.

• Supports one-click repair for MCP service connection exceptions.

• Optimized the output effect of the qwen3-thinking model in agent mode.

• Reduced the probability of flowchart generation under the code explanation instruction.

Fixes

• Fixed an occasional output compatibility issue when you switch between agent and file editing modes in project search.

Experience optimizations

• Optimized the effect of large file modifications in AI chat scenarios.

Feature updates

• New Inline Edit feature. You can invoke it by pressing `Ctrl`+`Shift`+`I` or `⌘`+`⇧`+`I`.

Fixes

• Fixed a delayed list update issue after a change in the MCP service's JSON file.

Fixes

• Fixed an issue where operation buttons were occasionally not visible during MCP tool execution.

• Fixed a display anomaly on the personal settings page on Windows systems.

Feature updates

• Tool use. The agent can autonomously use more than ten built-in programming tools, such as reading and writing files, code querying, and error troubleshooting. It also supports MCP tool configuration and integrates with the ModelScope MCP marketplace, which allows developers to configure freely.

• Terminal command execution. When executing coding tasks, the agent can autonomously decide which commands to execute, and automatically write and run commands in the terminal, which significantly improves the efficiency of coding task execution.

• Project-level changes. Based on the developer's task description, the agent can autonomously break down tasks and modify multiple code files within the project. It also supports gradual iteration or snapshot rollbacks through multiple conversations, which allows for collaborative coding with Lingma.

• Automatic project awareness. Based on the developer's task description, the agent can automatically perceive project information such as the framework, technology stack, required code files, and error messages. This eliminates the need to manually add project context and makes task descriptions easier.

• Memory awareness. The agent supports autonomous memory capabilities that are based on a large language model. As developers converse with Lingma, it gradually forms rich memories related to the developer, project, and issues, and becomes more knowledgeable with use.

Experience optimizations

• You can now navigate to files, methods, and definitions within the project from the AI chat.

• You can now add project directories as context in AI chat.

• You can now specify a knowledge base when you add context in AI chat (Enterprise Edition).

• You can now multi-select files, directories, and gitCommits when you add context in AI chat.

Feature updates

• You can now log on with an Alibaba Cloud International Website (www.alibabacloud.com) account.

Experience optimizations

• You can now use `Shift`+`Enter` for a line break in the Q&A window input box.

• Optimized the search box focus experience when you select context.

Fixes

• Removed the offline completion mode shortcut switch from the status bar.

Feature updates

• Supports logon for developers of the international exclusive edition.

Feature updates

• When you type in the Q&A window, it now recommends relevant code files based on the selected files.

• Supports configuring Project Rules.

• Supports one-click feedback for startup exceptions.

• When you use Qoder CN on the ModelScope platform, web logon is supported.

Experience optimizations

• Enhanced the performance of the Go language in inline code completion scenarios.

Fixes

• Fixed several bugs.

Feature updates

• The plugin now supports switching between Personal and Enterprise Edition member identities.

Experience optimizations

• Optimized the performance of inline completion for Java inner class calls.

• Optimized the similarity perception effect for inline completion.

• Optimized the automatic context import effect when you select input history in the input box.

• Optimized the perception of the entire file during Q&A when a code snippet is selected.

• Optimized the model response timeout issue.

Fixes

• Fixed an issue where model selection was not clickable in some scenarios.

Feature updates

• AI chat and AI Programmer now support model selection.

Experience optimizations

• Enhanced the historical memory of ongoing conversations for the AI Programmer.

• Optimized Mermaid language to not render graphs by default, with a one-click collapse option.

• Optimized for more concise answers in code explanation scenarios.

• Optimized the application speed of the AI Programmer for snippet modifications in large files.

• Optimized the file path for new files created by the AI Programmer.

Fixes

• Fixed an issue where image Q&A failed if the question was asked before the image was fully processed.

• Fixed an issue where completion was triggered multiple times when the cursor was in the middle of a tag.

Feature updates

• Supports one-click submission of Qoder CN plugin exception error logs.

Experience optimizations

• Optimized the experience of resolving startup issues, with one-click repair support.

• Optimized the effect of generating code files from images in AI Developer.

• Enhanced the SQL generation effect in code completion scenarios.

Feature updates

• Performance of optimized code for image generation in AI chat

• AI Programmer now supports generating code change files within the project from images.

• Enterprise exclusive edition accounts now support web logon in SSH Remote scenarios.

Fixes

• Fixed an issue where the language was occasionally not recognized when you generate files in an empty project.

Experience optimizations

• Continuously enhanced the cross-file awareness effect for Java and Go.

Experience optimizations

• Optimized the cross-file completion effect for Java and Go.

Feature updates

• Enhanced enterprise code generation now supports C/C++.

Fixes

• Fixed an issue where `.ipynb` files could not be selected when you choose context.

Feature updates

• AI chat now supports `codeChanges` and `gitCommit` contexts.

• Multi-modal image Q&A in AI chat now supports copying and pasting images.

• Simplified context selection for AI chat

Feature updates

• Supports configuring Project Rules.

• Added a feature to record model selections.

Fixes

• Fixed an issue where a line break in the Q&A window would overwrite the model selection area.

• Fixed an issue where the model might be reset.

• Fixed an issue where smart completion would overlap in some C++ projects.

Experience optimizations

• The AI chat window input box now supports using `Shift`+`Enter` for a line break.

Feature updates

• The AI chat window now supports model selection.

Fixes

• Fixed an issue that caused the Q&A window to send notifications.

• Fixed an issue where the Q&A status was incorrect when the right-click shortcut menu was used without selecting any code.

• Fixed some issues that could cause a crash.

• Fixed known issues with smart completion to improve the experience.

Feature updates

• You can now trigger "optimize code" and "generate unit tests" by typing `/` in the AI chat input or right-clicking in the editor area.

Experience optimizations

• Optimized the domain address input logic for the exclusive edition.

• Optimized the code block interaction in the AI chat window, which supports select all, highlight, and copy.

Feature updates

• You can now trigger "code explanation" and "comment generation" by typing `/` in the AI chat input or right-clicking in the editor area.

Fixes

• Fixed incorrect C# syntax highlighting in the AI chat window.

• Fixed an issue where the settings options panel failed to load.

Experience optimizations

• The AI chat window input now supports multi-line text input with line breaks.

Fixes

• Fixed issues in the completion scenario, such as parentheses not closing for "if" statements in certain situations.

Fixes

• Fixed an issue where the logon prompt bar did not appear in some cases.

Experience optimizations

• Optimized the dynamic response logic for the completion scenario.

Feature updates

• The AI chat window now supports displaying output links.

Fixes

• Fixed an issue where the session could not be stopped in the AI chat window.

Feature updates

• Supports syntax highlighting for the dark theme.

• The AI chat window now has highlighting for more languages, such as Markdown, SQL, and C++.

Feature updates

• Released the Qoder CN plugin for Visual Studio 2019, which supports:

• Real-time line-level/function-level code completion

• Code generation from natural language descriptions

• AI chat for R&D

Feature updates

• Added a settings window with logon/logoff support.

Fixes

• Fixed an issue where the manual trigger shortcut did not work in some scenarios.

• Fixed issues such as Markdown rendering problems.

Feature updates

• Supports customizing the keyboard shortcut for accepting inline code suggestions.

Fixes

• Fixed issues with inline code generation, such as occasional misplacement and editor event preemption by other plugins.

Feature updates

• Released the Qoder CN plugin for Visual Studio 2022, which supports:

• Real-time line-level/function-level code completion

• Code generation from natural language descriptions

• AI chat for R&D