The GitHub Copilot SDK's move from public preview to general availability is more than a version bump. It signals that GitHub is ready for teams to build production systems on the same agentic runtime powering Copilot — from internal developer tools and CI/CD assistants to customer-facing AI features. Since entering public preview, the SDK has already been used to build everything from release-note generators to support workflow agents.
SDK General Availability: What's Included
| Language | Install Command | Status at GA |
|---|---|---|
| Node.js / TypeScript | npm install @github/copilot-sdk |
Previously available |
| Python | pip install github-copilot-sdk |
Previously available |
| Go | go get github.com/github/copilot-sdk/go |
Previously available |
| .NET | dotnet add package GitHub.Copilot.SDK |
Previously available |
| Rust | cargo add github-copilot-sdk |
New at GA |
| Java | Maven and Gradle | New at GA |
Beyond new language support, the GA release brings several meaningful improvements based on preview feedback: better multi-client workflow support (multiple clients can contribute tools and permissions to the same session), slash commands and interactive input prompts now available across all SDKs, and improved diagnostics for debugging slow or failing connections. The API surface has been stabilized for production use.
The GitHub Copilot App: One Control Center for Agent-Native Development
The new GitHub Copilot desktop app introduces a "My Work" view that surfaces active sessions, issues, pull requests, and background automations across connected repositories in a single place. It is currently in technical preview for Copilot Pro, Pro+, Business, and Enterprise users.
Canvases: A bidirectional work surface for humans and agents. A canvas can show a plan, pull request, browser session, terminal, deployment status, or workflow state. Agents update canvases as they work; developers can edit, reorder, approve, or redirect work on the same surface without context switching.
Sandboxes: Choose whether Copilot runs on your local machine or in the cloud. Sandboxes give agents a bounded place to act while maintaining security policy enforcement and enterprise compliance — addressing one of the main objections to agentic workflows in regulated environments.
Copilot CLI Gets a Major Overhaul
GitHub Copilot CLI highlights at Build 2026
- Redesigned TUI in
/experimentalmode with tabbed access to pull requests, issues, and gists — staying in the terminal without losing repository context - Voice mode using on-device speech-to-text so audio never leaves the local machine
/everycommand for scheduling recurring prompts and background tasks- Cloud automations: agents run on a schedule or in response to GitHub events, opening issues and leaving comments (asks permission before write actions by default; autopilot mode available once trust is established)
- Memory++ and
/chronicle: Copilot maintains context across sessions started in the app, CLI, VS Code, or github.com — continuity across devices and over time - Partner agent apps from LaunchDarkly, Amplitude, Sonar, PagerDuty, Miro, and others integrate directly into the GitHub workflow
Platform Momentum Behind the Releases
GitHub's platform data contextualizes why these releases matter. Monthly commits have crossed 1.4 billion — nearly double year over year. GitHub Actions usage has exceeded 2 billion minutes per week. Repository creation, pull request activity, and API usage are all accelerating with no sign of plateauing.
The Copilot SDK GA and the new desktop app are GitHub's answer to a clear shift: developers are no longer building tools for humans to use manually. They are building agentic systems where AI does much of the work and humans maintain oversight, policy control, and final judgment. The same runtime that powers GitHub's own flagship product is now available to any team that wants to build on it.
For developers evaluating the agentic tooling landscape, the Copilot SDK's production-stable API, multi-language support, and deep integration with GitHub's existing repository and CI/CD infrastructure make it one of the most immediately usable options for embedding agent capabilities into real engineering workflows.