Welcome to Connext for GitHub Copilot

This Visual Studio Code extension brings the power of Connext AI to GitHub Copilot Chat.

To access the Connext expert, simply type @connext followed by your question in the Copilot chat window. Once you type @connext, the expert will stay active allowing you to ask multiple questions without having to type @connext again.

Requirements

To use the Connext expert, you need the following:

• An account on Connext AI
• A GitHub Copilot plan

To run Connext for GitHub Copilot, you need two additional extensions:

• GitHub Copilot - Provides inline coding suggestions as you type.
• GitHub Copilot Chat - A companion extension that provides conversational AI assistance.

These extensions are automatically installed as dependencies when you install Connext for GitHub Copilot.

In addition, some chat commands may require additional extensions. For example, the /createSystemDiagram command requires the Draw.io Integration.

Access

To use Connext for GitHub Copilot, log in to your Connext AI account by opening the command palette and selecting "Connext: Login".

You can log out of your Connext AI account by opening the command palette and selecting "Connext: Logout".

If you do not log in, the Connext expert will ask you to log in when you try to ask a question.

Features

Chat View: Get assistance from the Connext expert directly in the Copilot chat window by typing @connext followed by your question.

Quick chat: Initiate a brief conversation by opening Quick Chat with the "Chat: Open Quick Chat" command or by using the keyboard shortcut Ctrl+Shift+I or Cmd+Shift+I. This is useful for asking short, specific questions or getting quick help during coding—simply type @connext followed by your question in the chat window.

Context menu: You can obtain help from the Connext expert by right-clicking on a section of code, selecting "Connext" from the context menu, and choosing one of the available options.

Chat Commands

This extension allows invoking pre-defined commands from the chat window by typing @connext followed by the /<command>.

The available commands are:

• /showConnextInstallations: Displays information about the Connext installations (Micro ,Professional, or Drive) detected on your system. This command also allows you to set the default Connext installation and architecture used by the extension.
• /setConnextInstallation: Sets a Connext installation. The user will be prompted to select the directory where Connext (Micro, Professional, or Drive) is installed. This command does not select the installation as the default. To do that, use the /showConnextInstallations command. By default, the expert will try to detect the installation(s) directories automatically based on standard locations. However, if an installation is not detected, you can use this command to manually set the installation.
• /includeOpenFiles: The Connext expert will consider open files in the editor as part of the question context. To not type the command every time, you can toggle this setting on and off in the extension settings.
• /includeWorkspace: The Connext expert will consider all files in the workspace as part of the question context. To not type the command every time, you can toggle this setting on and off in the extension settings.
• /startAdminConsole: Opens RTI Admin Console using the default installation.
• /startSystemDesigner: Opens RTI System Designer using the default installation.
• /startMonitorUI: Opens the RTI Monitor UI using the default installation.
• /startShapesDemo: Starts the RTI Shapes Demo using the default installation.
• /quickStartExample: Generates a new example project using the default installation for a type and language of your choice. It also allows you to modify the publisher and subscriber code. This command uses the rtiddsgen tool to generate the example code. This command is designed to generate a simple, self-contained Hello-World type application — ideal for learning and experimentation, not for starting production projects.
• /createSystemDiagram: Opens the draw.io extension to create a diagram. From this diagram, you can generate a Connext system XML model using the command /generateSystemXmlModel. You can find an example diagram here.
• /generateSystemXmlModel: Generates a Connext system XML model from a draw.io diagram or an image (PNG, JPG, GIF, WebP). Open the file in Visual Studio Code and enter /generateSystemXmlModel. Example image here. If there are multiple files open, the command will use the first one.

The commands that start tools and generate new examples will fail if there is no default Connext installation configured (either by using the default installation directory or by setting the NDDSHOME environmental variable).

The commands that start tools may not work on Windows Subsystem Linux (WSL) depending on the OS and WSL versions.

Using natural language to invoke commands

You can use natural language to invoke commands, for example:

• @connext can you tell me what version of Connext am I running?
• @connext please start shapes demo
• @connext can you create a simple example for a basic sensor type in python? I would like the example to use waitsets in the subscriber

includeOpenFiles and includeWorkspace commands

The /includeOpenFiles and /includeWorkspace commands allow you to include open files and workspace files in the question context. Only files with the following extensions are considered: .xml, .drawio, .idl, .c, .h, .cxx, .cpp, .hpp, .py, .cs, .java, .txt, .md, .png, .jpg, .jpeg, .gif, .webp

Additionally, the following files automatically generated by rtiddsgen are ignored:

• For C: <IDL file name>Support.h, <IDL file name>Support.c, <IDL file name>Plugin.h, <IDL file name>Plugin.c, and <IDL file name>.c.
• Traditional C++: <IDL file name>Support.h, <IDL file name>Support.cxx, <IDL file name>Plugin.h, <IDL file name>Plugin.cxx, and <IDL file name>.cxx.
• Modern C++: <IDL file name>Plugin.hpp, <IDL file name>Plugin.cpp, and <IDL file name>.cpp.
• C#: Plugin.cs.
• Java: <type name>TypeCode.java, <type name>Support.java, <type name>DataReader.java, and <type name>DataWriter.java for each type in the IDL file.

The /includeWorkspace command also excludes the following directories: .git, .vscode, node_modules, CMakeFiles

New Example Command

The /quickStartExample command generates a new example project using the default installation for a specified type and language.

For example:

@connext /quickStartExample Temperature sensor in Python. The publisher should generate and publish temperature readings fluctuating between 60°F and 90°F.

@connext /quickStartExample Car battery state in modern C++. The subscriber must use a listener to receive the battery state.

This command produces a fully functional example, including all necessary files and code required for compilation and execution.

The command generates the example as follows:

1. First, it generates the IDL file based on the type description you provide.
2. Then, it generates the publisher and subscriber code based on the IDL file using rtiddsgen.
3. Then, it updates the publisher and subscriber code using the instructions you provide.
4. Finally, it generates the build system files and configuration files required to build, run, and debug the example.

To compile and run the generated code, you may need different extensions such as the CMake Tools extension.

C, Traditional C++, and Modern C++

For C, traditional C++, and modern C++, the /quickStartExample command generates:

• A CMakeLists.txt file for building the example.
• A CMakePresets.json file for configuring the build system.
• A "launch.json" file for running and debugging the example.

By default, the generator variable in CMakePresets.json is set to:

• "Unix Makefiles" on macOS and Linux.
• A generator derived from the default Connext architecture you chose on Windows after running the /showConnextInstallations command.

You can override the generator by configuring the cmake.generator option if the default generator is not suitable or available in your environment.

To list the available generators, run the following command in the terminal:

cmake --help

Java

For Java, the /quickStartExample command generates:

• A setting.json file for building the example.
• A "launch.json" file for running and debugging the example.

Python

For Python, the /quickStartExample command generates:

• A "launch.json" file for debugging the example.

The Python example requires installing the Connext Python API. For details on how to install the Connext Python API, refer to the RTI Connext DDS Python API Getting Started Guide.

C# and .NET

For C# and .NET, the /quickStartExample command generates:

• A "launch.json" file for running and debugging the example.

License

Copyright (c) 2025 Real-Time Innovations, Inc. All rights reserved.

RTI grants you the right to compile and use this software. You may not modify, create derivative works of, or distribute the software.

This software is an experimental (i.e., pre-production or beta) product. It is provided “as is”, with no warranty of any type, including any warranty for fitness for any purpose. RTI is under no obligation to maintain or support the Software. RTI shall not be liable for any incidental or consequential damages arising out of the use or inability to use the software.

This software integrates with AI services provided by RTI, and your use of those services is subject to RTI's Privacy Policy and these Terms of Use.

Please provide feedback via email to the Connext AI team at connext-ai-feedback@rti.com.

For further assistance or inquiries, you can also reach out to RTI's support team at support@rti.com.