Weekly GitHub Copilot Roundup: SDK Agents, CLI Plan Mode, Modernization
GitHub Copilot continues to add new agent-based SDK capabilities, as well as improvements for its command-line interface and tools for automating modernization. Developers can now use Copilot in different development environments, including IDEs, terminals, automated pipelines, and custom agent-based apps. The latest enhancements to the CLI, better integration with organizational information, and tools for Java modernization are all designed to provide more concrete benefits for programming teams.
GitHub Copilot SDK: Powering Custom Agentic Development
The GitHub Copilot SDK is now in technical preview, following a recent rollout for platforms including Node.js, Python, Go, and .NET. Step-by-step guides now demonstrate how to use Copilot’s code completions inside purpose-built agent applications and automation tasks. The SDK expands on CLI capabilities with features such as agentic workflows, memory, tool orchestration, routing for multiple models, and streaming. Microsoft and contributors are offering in-depth examples for embedding agents using Python, enabling .NET in Visual Studio, and combining Copilot with external model authentication. The platform’s modular approach is suitable for organizations creating custom AI-enabled workflows and tools, and user feedback notes broader integration with internal processes.
- Bringing Work Context to Your Code with GitHub Copilot SDK
- Building Agentic Apps with the GitHub Copilot SDK: New Developer Paradigms
- Build an Agent into Any App with the GitHub Copilot SDK
- Using the GitHub Copilot SDK with Python
- The GitHub Copilot SDK is here! - Rubber Duck Thursdays
- Open Source Friday: Exploring the GitHub Copilot SDK
- Add an AI Agent to Your Application with GitHub Copilot SDK
GitHub Copilot CLI: Enhanced Terminal Workflows and Integration
GitHub Copilot CLI now offers new features for automatically using AI within the terminal, following updates such as support for different models (including GPT-5 mini, GPT-4.1), better installation, and agent-driven command-line tools. The updated “Plan” mode introduces guided, stepwise code planning before code is generated, which aligns with the broader move toward agentic workflows. GPT-5.2-Codex support provides improved prompt handling and context control (/context). The CLI now integrates with gh copilot for quick setup and unified onboarding for teams. Features for background task delivery (/delegate), persistent repository memory, review functionality, and audit/history tools all support better security practices. The YOLO mode and CI/CD automation are designed to cut repeated tasks.
- Install and Use GitHub Copilot CLI Directly from the GitHub CLI
- GitHub Copilot CLI: Plan Mode, Advanced Reasoning, and Terminal Workflow Enhancements
- A Cheat Sheet to Slash Commands in GitHub Copilot CLI
- Demo: Using /delegate in the GitHub Copilot CLI
- Demo: Using GitHub Copilot CLI and YOLO Mode
- Building with GitHub Copilot CLI: Rubber Duck Thursdays Live Coding Stream
Organizational and Work Context Integration
For distributed and enterprise development teams, Copilot now adds greater organizational context into AI features. Following last week’s update on context engineering and memory, these integrations allow Copilot to pull from Microsoft 365, SharePoint, and call transcripts via the Work IQ MCP server. You can now search organization documents and previous work history directly from the IDE or command line, which helps with requirement matching and audit needs. Persistent and auditable context handling strengthens code reviews and onboarding, which continues the trend of team-wide curation and verifiable memory functions. Teams report higher productivity and more accurate requirement handling.
- Bringing Work Context to Your Code with GitHub Copilot SDK
- Bringing Organizational Context to GitHub Copilot CLI with Work IQ
- Bringing Work Context to Your Code in GitHub Copilot
App Modernization with GitHub Copilot
Copilot’s modernization tools support more automation for Java and Spring upgrades. Last week’s Java EE to Jakarta EE migration is now followed by tools to automate updates for Spring Boot and Spring Framework, with step-by-step guides for both JDK upgrades and secure identity changes. These tools combine security, dependency management, and automated refactoring—including support for OpenRewrite, JDK/build planning, and Microsoft’s ID and Key Vault. This reduces friction for enterprise migrations, covering tasks like namespace changes, dependency and CVE analysis, fixing security vulnerabilities, and deeper IDE integration. Focus remains on simplifying post-migration review, automated code improvements, and cloud adoption.
- Modernizing Spring Boot Applications with GitHub Copilot App Modernization
- Modernizing Applications by Migrating Code to Managed Identity with GitHub Copilot App Modernization
- Migrating Application Credentials to Azure Key Vault with GitHub Copilot App Modernization
- Modernizing Spring Framework Applications with GitHub Copilot App Modernization
- Upgrade Your Java JDK (8, 11, 17, 21, or 25) with GitHub Copilot App Modernization
Integrating and Expanding Copilot Use Across Platforms
GitHub Copilot continues to integrate with additional workflows. After last week’s news about Copilot support for OpenCode (including login and credential management across desktop, terminal, and IDE), current guides show practical setup steps for connecting Copilot to OpenCode and CI/CD tools. An in-depth look at Copilot’s Arm Cloud Migration Agent discusses how Copilot helps with container migration, indicating expanded applicability in infrastructure and cloud migrations. Community updates show more widespread adoption.
- How to use GitHub Copilot with OpenCode
- GitHub Copilot Arm Cloud Migration Agent Deep Dive
- The Download: GitHub Copilot SDK Updates, Copilot for OpenCode, and Cloudflare Buys Astro
Advanced Workflow Guidance and Tutorials
Recent resources offer strategies for using Copilot throughout the development process. Tutorials cover test-driven development with agents, Copilot review practices, and ways Copilot functions as a collaborator rather than just a code assistant. KQL support in Microsoft Fabric continues to show how Copilot can power analytics and engineering. Community events like the .NET AI Community Standup discuss new SDK features and integration, highlighting practical implementation advice.