Weekly GitHub Copilot Roundup: Open VS Code, More Models

by TechHub

GitHub Copilot received new updates, including open sourcing of its VS Code extension, expanded control over AI models, and integrations focused on transparency for developers. Automation, security, and configuration options now better support individual and enterprise needs. Tutorials highlight Copilot's role in telecom APIs, certification study, and open source development. Copilot continues to evolve as a flexible AI coding platform, assisted by community contributions and support for additional models.

VS Code Copilot Extension: Open Source, Custom Modes, and Deep Customization

Following last week’s focus on MCP integration and workflow automation, the VS Code Copilot Extension is now open source, discussed in detail by Microsoft’s Burke Holland. Community access allows developers to contribute new features and refine daily Copilot usage in VS Code. Custom Modes support tailored AI responses and performance for specific developer workflows. Advanced users can enable ‘Beast Mode’ in VS Code Insiders to experiment with prompt engineering and agent tuning. These improvements build on last week’s MCP-driven automation. Guides and documentation on Custom Modes offer hands-on prompt engineering instructions. Open sourcing welcomes contributions such as code fixes and feature suggestions, letting users directly affect Copilot’s development. Initial coverage explains the extensibility available for customizing Copilot behavior.

GitHub Copilot in Visual Studio Code: New Features, AI Models, Security, and Workflow Enhancements

After agent and prompt file improvements last week, Visual Studio Code August 2025 (v1.104) introduces more Copilot upgrades for streamlined coding and improved security. Copilot Chat now automatically selects among AI models (GPT-5, Claude Sonnet 4, Gemini Pro 2.5), enhancing suggestion accuracy and dealing with rate limits; manual model selection is also possible. Team collaboration gets refinements such as an advanced agent system, reusable prompts, device-wide toggles, and chat enhancements including KaTeX math rendering and custom fonts. Extension authors benefit from finalized LanguageModelChatProvider APIs and improved authentication. For security, agent mode prompts users before modifying protected files, and organizations can adjust approvals for terminal commands. Changed files are now easier to review with collapsible lists, and expanded AGENTS.md support allows detailed Copilot usage standards. These additions align with productivity and high-control needs for large and security-sensitive teams.

Expanded AI Model Support and BYOK Flexibility

Extending last week’s updates for organizational customization, Copilot now supports OpenAI GPT-5 and GPT-5 mini—GPT-5 mini for Free users, GPT-5 for paid subscriptions. A new model picker UI lets developers and organizations choose models on GitHub.com, VS Code, Visual Studio, JetBrains IDEs, Xcode, and GitHub Mobile. IT administrators can set defaults centrally. This lets users and teams adjust Copilot for anything from basic scripting to complex code generation. Copilot now supports Bring Your Own Key (BYOK), available in public preview for JetBrains IDEs and Xcode. Developers can use their own API keys for Azure, OpenAI, Anthropic, Gemini, Groq, and OpenRouter—enabling model experimentation and better quota management. Teams can customize Copilot Chat and try out new models before they’re fully supported, with ongoing community feedback shaping future IDE integrations.

GitHub Copilot Coding Agent and Agentic Workflows

Building on agent-driven automation and MCP connections from last week, Copilot’s agent now acts as a coding teammate—running code fixes, refactoring, and other pull request tasks. Developers assign agent tasks via Issues, VS Code, or a special panel, using ephemeral GitHub Actions runners for safe, isolated work. Agents’ changes are reviewed and run through CI before approval, maintaining secure standards. The Model Context Protocol (MCP) provides expanded context for improved results. Teams get management over assignments, audits, and branch protection, with guides outlining setup and scaling.

Automatic Code Review and Administration with Copilot and MCP

Copilot’s new repository rule allows separate automatic code review from merge requirements. Repository admins can get early feedback and frequent code quality checks tailored to different rule sets, supporting productivity and flexible coverage—useful for regulated projects. The MCP remote server (now generally available) links Copilot agents, GitHub workflows, and large language models. Security features include secret scanning, code scanning, and automated advisories. Enterprises can set internal registry and allowlist controls for MCP servers via Copilot in VS Code Insiders, regulating AI endpoint access. These changes make configuration easier and anticipate broader tool support, showing Copilot’s focus on balancing useful AI and governance.

Copilot for TM Forum Open API and Telco Workflows

Copilot’s application in TM Forum Open API development boosts productivity and standards compliance for telecom APIs—especially with Node.js/Express TMF-compliant endpoints. Copilot streamlines boilerplate creation, testing, and validation. Companies like Proximus, NOS, Orange, and Vodafone report faster development and improved API matching. Best practices stress ongoing validation and keeping features current. At TM Forum Innovate Americas 2025, Microsoft demonstrated Copilot, Azure AI Foundry, and MCP integration for modular telecom architectures and open standards. Agentic AI and Copilot accelerate API delivery, reduce repetitive work, and support orchestration, showing impact in telecom engineering.

Tutorials and Practical Guides: Prompt-Driven Development, Building with AI, and Certification Support

A range of resources are now available. Tutorials in VS Code detail Copilot prompt engineering, Copilot Vision, voice interactions, and agent workflows—helping users automate coding and boost test coverage from within the IDE. Building on last week’s beginner guides and prompt best practices, other guides cover Copilot’s use in open source app prototyping and Microsoft certification study, offering step-by-step automation for documentation, scripting, and custom workflows (like Markdown or PowerShell tasks for cloud automation). Copilot helps developers build skills, onboard to new stacks, and launch open source projects with community support.

Other GitHub Copilot News

GitHub Universe 2025 will feature over 100 sessions about Copilot workflows, automation, and current advances in AI and security. Attendees can view demos, take remote certifications (including new Copilot exams), and meet technical experts, showing Copilot’s growing role in development and automation.