Visual Studio Insiders release notes

JSON editor updated to support more JSON schema features.

The JSON editor now supports newer JSON Schema specifications, giving you access to modern schema features like $defs, $anchor, and improved vocabulary support.

Previously, the editor only supported JSON Schema Draft 4 and Draft 7. You can now use:

• JSON Schema Draft 2019-09
• JSON Schema Draft 2020-12

This means better IntelliSense and validation when working with schemas that use the latest specifications.

Copilot agents in Visual Studio now automatically discover and use skills defined in your repository or user profile.

Copilot agents in Visual Studio can now automatically discover skills. Agent Skills are reusable sets of instructions that teach agents how to perform specific tasks, like running a build pipeline, generating boilerplate, or following your team's coding standards. By defining skills once, you save time and ensure consistent behavior across your workflow. See awesome-copilot for example skills shared by the community.

Skills are picked up from the following locations:

Skill type	Location
Workspace or project skills (stored in your repository)	.github/skills/, .claude/skills/, .agents/skills/
Personal skills (stored in your user profile)	~/.copilot/skills/, ~/.claude/skills/, ~/.agents/skills/

To create a skill:

1. Create a .github/skills/ directory for a workspace skill or a ~/.github/skills/ directory for a personal skill.
2. Create a subdirectory for your skill. Each skill should have its own directory (for example, .github/skills/github-issues/).
3. Create a SKILL.md file in the skill directory. For the format of SKILL.md, follow the instructions on agentskills.io/specification.
4. Optionally, add scripts, examples, or other resources to your skill's directory.

Once you create a skill in one of these directories, the agent will decide when to use it as appropriate. When a skill is activated, it appears in the chat to let you know it's being applied.

Note that more user-friendly flows for browsing and creating skills inside Visual Studio are coming in the next few releases. Please give Agent Skills a try and let us know your feedback!

📣 See feature ticket to share your feedback and continue the conversation.

Start new cloud agent sessions directly from Visual Studio.

Cloud agents, like Copilot coding agent, run on remote infrastructure for scalable, isolated execution. We are excited to share that now you can initiate new cloud agent sessions directly from Visual Studio!

To try out the cloud agent, simply click Cloud from the agent picker in the Chat window and then share the work you'd like it to assist with. Cloud agent will first ask your permission to open an issue for the task in your repository, and then create a pull request to address it. While cloud agent is completing the PR, you can feel free to parallel working on other items in VS, or even turn off VS and come back later. Once the PR is ready, you will be notified and you can review it inside VS or on browser.

Note that cloud agent in VS is currently powered by Copilot coding agent. To use it, make sure you are in a GitHub repository and that Copilot has permission to create issues in that repo. Please give cloud agent a try today and let us know your feedback!

Define custom agents tailored to your team's workflow using .agent.md files in your repository or user profile.

Custom agents let you build specialized Copilot agents tailored to your team's workflow. They have full access to workspace awareness, code understanding, tools, your preferred model, and MCP connections to external knowledge sources like internal documentation, design systems, APIs, and databases.

Custom agents are defined as .agent.md files in your repository or your user profile. Workspace or project-specific agents are stored in your repository under .github/agents/. For example:

your-repo/
└── .github/
    └── agents/
        └── code-reviewer.agent.md

User-level agents are stored in %USERPROFILE%/.github/agents/ by default, and the location is customizable in Tools → Options → Copilot Chat → Custom agents user directory.

To create a custom agent, you can either create the .agent.md files directly at the locations above, or click the + button from the agent picker and follow the steps from there.

Successfully created custom agents will appear in the agent picker for your use at any time.

A few things to note:

• If you don't specify a model, the agent uses whatever is selected in the model picker.
• Tool names vary across GitHub Copilot platforms - check the tools available in Visual Studio to make sure your agent works as expected.
• The awesome-copilot repo has community-contributed agent configurations you can use as starting points. Verify tool names before using them in VS.

📣 See feature ticket to share your feedback and continue the conversation.

Customize the keyboard shortcuts for accepting Copilot inline suggestions.

We are excited to share that you can now customize the keyboard shortcuts for accepting Copilot inline suggestions. Whether you want to change the key for accepting a full suggestion, the next word, or the next line, it's all configurable through the standard keyboard settings.

To set up your preferred shortcuts:

1. Go to Tools → Options → Environment → Keyboard
2. Search for the command you want to customize:
   • Edit.AcceptSuggestion
   • Edit.AcceptNextWordinSuggestion
   • Edit.AcceptNextLineinSuggestion
3. Remove the existing keybinding
4. Assign a new shortcut under the Inline Suggestion Active scope

For example, in the image below, the default accept shortcut is changed from Tab to Ctrl+Tab. The updated shortcut is reflected throughout the hint bar.

📣 See feature ticket to share your feedback and continue the conversation.

IntelliSense now takes priority over Copilot completions so you only see one suggestion at a time.

We heard your feedback that it can be confusing when both IntelliSense and Copilot completions appear at the same time. To address this, the editor now prioritizes IntelliSense list over Copilot and only shows one suggestion at a time.

When IntelliSense is active, Copilot completions are temporarily suppressed so you can focus on the IntelliSense list selection without distraction. Once you dismiss or commit the IntelliSense selection, Copilot completions resume as normal. This is now the default behavior. Update your VS to the latest version and code as you normally do. Please let us know any feedback you have!

📣 See feature ticket to share your feedback and continue the conversation.

Browse and navigate your Copilot chat sessions with the new chat history panel.

The chat history experience has been redesigned from a simple dropdown list to a dedicated panel. Each entry now displays the chat title, a preview of the last message, and when the session was last updated, making it easier to find and jump between past conversations.

More capabilities for managing chat history are on the roadmap. We'd love to hear what you want to see. Please file a ticket on Developer Community to share your feedback!

📣 See feature ticket to share your feedback and continue the conversation.

Agentic Bug Resolution helps you find and fix bugs with runtime-proofed, interactive AI guidance.

We are introducing a new Debugger Agent workflow in Visual Studio that validates bugs against real runtime behavior instead of relying on static analysis guesses. The experience walks you through a complete agentic loop from understanding and reproducing the issue, to instrumenting the application, isolating the root cause, and validating the fix through live execution.

You can start from an issue in GitHub or Azure DevOps, or simply describe the bug or code behavior in natural language. The Debugger Agent uses this input to map the problem directly to your local source and guide you through resolution. Unlike static approaches, you can interact with the agent during the debugging process to provide more input, discuss your theory, or refine the fix in real time.

How the agentic loop works:

• Context Injection – Provide an issue link or describe the bug in Debugger Mode in Copilot Chat (you can switch modes using the dropdown in the lower-left corner of the chat). The agent connects the report to your local code.
• Autonomous Reproducer – The agent analyzes the bug description and, if reproduction steps are missing, creates a minimal scenario to trigger the failure automatically.
• Hypothesis & Instrumentation – It generates failure hypotheses and instruments your app with tracepoints and conditional breakpoints to capture runtime state.
• Runtime Validation – The agent runs the debug session and analyzes live telemetry to isolate the root cause.
• Targeted Correction – It suggests a precise fix at the exact failure point rather than broad refactoring.
• Final Human Validation – You rerun the scenario and confirm the fix in the live environment alongside the agent.
  

C++ Code Editing Tools for GitHub Copilot Agent Mode are now available by default.

We're excited to share that C++ Code Editing Tools for GitHub Copilot Agent Mode are now generally available by default. These tools help Copilot navigate your C++ codebase more effectively when refactoring or modifying code.

With these tools enabled, Copilot can:

• Map out class inheritance hierarchies
• Follow function call chains

To get started:

1. Open a C++ project and ensure IntelliSense is configured.
2. Enable specific tools using the Tools icon in Copilot Chat.

Once enabled, Copilot uses these tools automatically when answering questions or suggesting code changes.

For best results, use AI models that support tool-calling. Learn more at AI model comparison - GitHub Docs