The VS Code Live stream for the v1.102 release highlighted new open source initiatives, AI integrations like the Copilot Chat extension and Copilot Coding Agent, and enhanced user experiences including Model Context Protocol support and customization features. The team emphasized community involvement, transparency, and upcoming opportunities for developers to engage with these innovations through the VS Code Insiders build and events like MCP Dev Days.
The VS Code Live stream for the v1.102 release showcased several exciting new features and improvements, with a focus on open source contributions, AI integrations, and enhanced user experiences. Olivia, the host, introduced the stream by highlighting the importance of the monthly release live streams as a platform for the VS Code engineering and product management teams to demo new features and engage with the community. She emphasized the open source nature of VS Code and the team’s commitment to transparency, especially with the recent open sourcing of the Copilot Chat extension, which Izzy detailed in the first segment.
Izzy discussed the open sourcing of the VS Code Copilot Chat extension, explaining how users can now explore the AI features directly on GitHub, understand the prompting mechanisms, and even troubleshoot issues using the new chat debug view. This move aligns with the team’s long-standing passion for open source, enhancing transparency around AI functionalities and data collection. Izzy also highlighted the community’s role in contributing to the AI experience, encouraging users to file issues and submit pull requests to help evolve the project further.
Sandeep then presented the general availability of the Model Context Protocol (MCP) support in VS Code, showcasing a new user-friendly UI that simplifies discovering, installing, and managing MCP servers. This update makes MCP more accessible to users by mirroring the familiar extension installation experience and providing controls to start, stop, and monitor MCP servers within the editor. Sandeep also touched on enterprise features like policy controls for MCP enablement and improvements in handling API keys and authentication flows.
Josh demonstrated the integration of the Copilot Coding Agent within VS Code, sharing a personal project where the agent autonomously handles coding tasks, bug fixes, and feature implementations. He showcased how the agent can be delegated tasks, work asynchronously, and produce pull requests that developers can review and merge. Josh also highlighted the use of local MCP servers to enhance the agent’s capabilities by providing up-to-date documentation and context, improving the quality of AI-generated code, especially for specialized languages like Swift.
Harold and Justin rounded out the stream with demos on customization and user experience enhancements. Harold introduced a community-driven repository for custom instructions, reusable prompts, and chat modes that help tailor AI workflows to specific development needs, including advanced planning and testing strategies. Justin showcased new experimental features like editing previous AI requests and checkpoints for undoing and redoing changes, which improve the interactivity and control developers have over AI-assisted coding sessions. The stream concluded with a call to action for viewers to try out these features in the VS Code Insiders build and participate in upcoming events like MCP Dev Days to deepen their engagement with the evolving AI and extension ecosystem.