VS Code Insiders Podcast

VS Code's Agent Host Protocol Explained with Connor Peet

VS Code’s Connor breaks down the shift from local, window-bound agents to a more flexible Agent Host Protocol that can keep sessions alive, sync state across machines, and support remote, long-running workflows. He explains how the protocol opens the door to richer orchestration, multi-client experiences, and custom tools built on top of VS Code’s agent ecosystem—while hinting at where agent-powered development is headed next.

• Transcript
• Chapters

Agent-First Development Workflows in VS Code with Brigit Murtaugh

Explore VS Code’s new Agents window — a familiar, focused UI that puts agent-first workflows front and center across repos, harnesses, and machines. It centralizes session management with worktrees, integrated previews, diffs, run tasks and PR flows so agents can iterate across projects while you review, customize extensions, and switch to the editor to edit.

• Transcript
• Chapters

Inside The Agent Loop with Pierce Boggan

Dive into the heart of VS Code’s agent loop as James and Pierce unravel how agents, sub-agents, tools, and harnesses work together to power your coding experience. Learn how the agent loop has evolved, why different models are used for specific tasks, and how you can customize or optimize your own workflows. Perfect for anyone curious about the magic behind AI-driven code assistance!

• Transcript
• Chapters

Autopilot Mode with Justin Chen

Justin and James deep-dive into Autopilot and the evolving VS Code chat UX—why shimmers and collapsed containers declutter conversations, and why the input bar split and new permissions picker matter. Learn how Autopilot (Insiders preview) can auto-approve tools, answer prompts, iterate until a task_complete signal or max retries, and when to use default vs bypass approvals; practical tips for safe, hands-off workflows and feedback.

• Transcript
• Chapters

Subagents: Parallel Execution and Context Isolation [Updated]

Discover how VS Code's subagents revolutionize parallel execution and context management. Learn the fundamentals of agent loops and context windows, then dive into how subagents isolate tasks, run simultaneously, and keep your main agent focused. Plus, explore custom agents and model optimization strategies to supercharge your development workflow.

• Transcript
• Chapters

Orchestrating Multiple Agents in VS Code with Ben & Peng

In this episode James sits down with Ben and Peng from the VS Code team to unpack a year of AI evolution inside the editor — from Copilot Free to the new agent sessions UI. They explain the three agent flavors (interactive local, background, cloud), why agent orchestration matters, and how VS Code is reshaping UX so you can run parallel sessions, delegate work, and treat agent outputs like inbox items to triage. Key takeaways: build context in local chats then delegate to background/cloud, harde

• Transcript
• Chapters

VS Code - 2025 Wrapped with Burke Holland and Pierce Boggan

Burke and Pierce sit down to review a full year of releases, surprises, partnerships, and AI goodies in VS Code from 2025!

• Transcript
• Chapters

Models, Evals, and Raptor Mini with Julia Kasper

In this episode of the VS Code Insiders podcast, James sits down with Julia to explore the fast-changing world of AI models inside VS Code. They dive into how developers can evaluate models, the role of fine-tuning, and why speed versus complexity matters. From Raptor Mini to Sonnet, this conversation reveals the behind-the-scenes work shaping your coding experience today.

• Transcript
• Chapters

Taming AI Assisted Coding Models with Eleanor Berger

Pamela and Eleanor dig into the real art of AI‑assisted coding—delegating work to background agents, writing specs that models actually follow, and steering different LLM personalities without superstition. Expect hands‑on tactics with Copilot, GitHub Actions, CLI‑first workflows, and the agents.md standard, plus a peek at Eleanor’s open‑source Ruler project for centralized rules. Come for the model face‑offs, stay for the practical tips: richer prompts, sandboxed evals, and even talking to your

• Transcript

Designing for Everyone: Accessibility in VS Code with Megan Rogge

Megan Rogge, Software Engineer on VS Code, joins James to explore how accessibility is woven into the culture and engineering practices of Visual Studio Code. From screen reader support to sound signals and contrast themes, this episode reveals the thoughtful design behind features that empower every developer.

• Transcript
• Chapters

Beyond the Keynote: VS Code at GitHub Universe 2025

What does it look like for VS Code to be an AI-native editor? How is it evolving with agents running locally in the editor, in a CLI, in the cloud, or even from other companies? Pierce breaks down every part of what was announced for VS Code at GitHub Universe and goes behind the scenes of his part of the keynote.

• Transcript
• Chapters

Behind the Scenes of VS Code’s Planning Agent with Bhavya U

The new planning agent is here to help you plan out your next tasks, but how does it differ from todos? When would you want to use the planning agent, and how does it actually work? Bhavya, software engineer on the VS Code team, sits down to break it all down.

Building & Scaling Open Source Communities with Steve Francia

Steve Francia joins the pod to talk about building, scaling, and supporting large open source projects and communities from MongoDB, Go, Cobra, Hugo, and so much more.

Mastering Chat Modes in VS Code with Burke Holland

Burke joins the pod to talk about chat modes, how they are evolving, how you can use them as agents, and how he created Beast mode for GPT 4.1 that revolutionized working with this model.

Building agent memory for VS Code with Harald Kirschner

Harald joins James to talk through customizing chat, agents, and GitHub Copilot with custom instructions, prompts, and modes. We go deeper int new proposed APIs that pulls them together to create agents and how Harald implement agent memory and shipped it as an extension!

Impact of AI on How We Learn to Code with Scott Tolinski

James sits down with Scott Tolinski, of Syntax podcast, to talk about how developer education has changed over the years since he started creating tutorials, how AI is impacting how we learn to code and problem solve, and what the future looks like.

The Origins & Evolution of the GitHub MCP Registry

Toby sits down to talk about MCP, model context protocol, why it is important, how the steering commit across companies helps form the open standard, and how the MCP registry standard came to be and how GitHub and VS Code are leveraging it.

The Future of Coding Agents in VS Code

Brigit and Josh join the podcast to talk about the latest enhancements to coding agents, aka background tasks, aka background agents in VS Code. They break down why you would use them, how to use them, and their future.

Life of a VS Code SWE Intern with Christy Nguyen

Christy joins us to talk about her internship on the VS Code team, her journey, how she landed the internship, how she developed git worktree support in VS Code, and what's next for her.

Language Model Chat Provider (aka BYOK) API with Logan Ramos

This week, VS Code engineer Logan Ramos joins the pod to talk about Bring Your Own Key... or Bring Your Own Model... the ability to configure models from several providers including Azure, OpenAI, xAI, Anthropic, Google, and more. We talk about how and why you would actually use this functionality, but more importantly a new API for extenders to bring ANY model directly into VS Code!