The VS Code Team shares how the GitHub Copilot Chat extension now unifies all Copilot AI code suggestions in a single open source extension, with technical details and contribution opportunities.

Open Source AI Editor: Second Milestone

By The VS Code Team

Introduction

Visual Studio Code (VS Code) is advancing its journey toward becoming a fully open source AI-powered code editor. As of September’s Version 1.105, the Copilot Chat extension now delivers all inline AI code suggestions—previously split between two extensions—in a unified, open source package. This marks the second major milestone in their open source AI initiative.

Consolidation of Copilot Features

  • Single Extension Experience: All Copilot functionality—inline suggestions, ghost text, chat, and agent mode—is now available via the Copilot Chat extension.
  • Deprecation Plan: The old GitHub Copilot extension (previously responsible for inline ghost text) will be deprecated by early 2026 and removed from the VS Code Marketplace.
  • Transparent Transition: Existing users should experience no major changes. Inline suggestions and chat features have been ported to the new unified extension with ongoing testing and progressive rollout.

Technical Overview: How Inline Suggestions Work

Developers can now explore and contribute to the vscode-copilot-chat repository. Here’s how the inline suggestion pipeline operates:

  1. Typing-as-suggested Detection: Checks if the user is following a previous AI suggestion to avoid unnecessary computation.
  2. Caching: Utilizes cached suggestions for performance, reducing repeated calls to language models.
  3. Reusing Ongoing Requests: Leverages still-running LLM requests when possible to minimize latency.
  4. Prompt Construction: Gathers relevant code and context to formulate prompts for the LLM.
  5. Model Inference: Requests ghost text and next edit suggestions, prioritizing what’s available at the cursor location.
  6. Post-processing: Refines raw outputs for correct code style, indentation, and syntax.
  7. Multi-line Intelligence: Determines how many lines to suggest based on context and confidence.

Performance Improvements

  • Reduced Latency: Networking optimizations provide faster AI code completions.
  • Quality Validation: Rigorous experiments ensure no drop in performance or suggestion relevance during the migration.

Community Involvement

  • Open Contributions: The extension’s codebase invites community PRs and bug reports through its GitHub repository.
  • Iteration Plans: Progress and upcoming work are available via iteration plans.
  • Troubleshooting Options: Users uncomfortable with the new unified setup can temporarily revert to prior extension behaviors.

Future Direction

The next phase will refactor core AI functionality directly into VS Code’s core, increasing extensibility and community involvement. Developers can expect ongoing improvements in AI-based development workflows within VS Code.

Explore, contribute, and follow along with VS Code’s open source AI journey!

This post appeared first on “Visual Studio Code Releases”. Read the entire article here