The past week saw GitHub Copilot become a built‑in extension in Visual Studio Code 1.116, reducing onboarding friction for new users. At the same time, GitHub announced significant changes to individual Copilot plans, pausing new sign‑ups and tightening usage limits as agentic workloads increase compute demands. Claude Opus 4.7 rolled out across Pro+, Business, and Enterprise tiers, delivering stronger multi‑step reasoning for complex coding tasks.
VS Code 1.116 Makes Copilot a Built‑in Extension
With the release of Visual Studio Code 1.116 on April 15, GitHub Copilot Chat is now a built‑in extension. New users no longer need to install any extension to start using chat, inline suggestions, and agent features. Existing users are unaffected, and AI features can still be disabled via the chat.disableAIFeatures setting. This change streamlines the onboarding experience and further integrates AI‑powered capabilities into the core editor.
Why it matters for enterprise teams: Reducing setup steps lowers the barrier to adoption for developers who are evaluating Copilot. For organizations rolling out Copilot at scale, the built‑in presence reinforces Copilot as a first‑class component of the development environment, not an optional add‑on.
GitHub Tightens Individual Copilot Plans, Citing Agentic Workloads
On April 20, GitHub announced a series of changes to Copilot Individual plans (Pro, Pro+, and Student). New sign‑ups for these plans are paused, while existing subscribers face tighter usage limits. Opus models have been removed from the Pro tier—Opus 4.7 remains exclusive to Pro+—and the Pro+ tier now offers over five times the usage limits of Pro. The company noted that “agentic workflows have fundamentally changed Copilot’s compute demands,” reflecting the resource intensity of long‑running, parallelized agent sessions.
Why it matters for enterprise teams: The shift underscores the growing cost of agent‑based coding assistance. Organizations that rely on Copilot for heavy automation should monitor their usage patterns and consider upgrading to Business or Enterprise plans where usage is billed per user, not per token. The changes also signal that GitHub is adjusting its pricing model to better align with actual compute consumption.
Claude Opus 4.7 Generally Available with Stronger Agentic Execution
Claude Opus 4.7, Anthropic’s latest flagship model, began rolling out to GitHub Copilot on April 16. It replaces Opus 4.5 and 4.6 in the model picker for Copilot Pro+, Business, and Enterprise users. Early testing shows improved multi‑step task performance and more reliable agentic execution, making it better suited for complex refactoring, debugging, and tool‑dependent workflows.
Why it matters for enterprise teams: Teams that depend on Copilot for sophisticated, multi‑file changes will see more consistent results and fewer “hallucinations.” The model’s stronger reasoning capabilities can reduce the need for manual intervention, accelerating development cycles for large‑scale projects.
Enable Copilot Cloud Agent via Custom Properties (Enterprise)
Enterprise administrators can now selectively enable GitHub Copilot Cloud Agent (CCA) access on a per‑organization basis using custom properties. Previously, admins could only enable or disable the agent globally. The new granular control, announced on April 15, allows pilot programs with specific teams and gradual rollouts across large organizations.
Why it matters for enterprise teams: Fine‑grained policy management reduces risk when introducing agent‑based automation. Security‑conscious organizations can start with a limited group of trusted developers, monitor usage and outcomes, and expand access gradually without exposing the entire organization.
Manage Agent Skills with GitHub CLI (gh skill)
GitHub launched gh skill, a new CLI command in GitHub CLI v2.90.0+, on April 16. The command lets developers discover, install, manage, and publish agent skills that work across multiple AI coding agents, including GitHub Copilot, Claude Code, Cursor, Codex, and Gemini CLI. Skills follow the open Agent Skills specification and support version pinning, content‑addressed change detection, and immutable releases for supply‑chain integrity.
Why it matters for enterprise teams: Standardized skill management enables teams to codify best practices—such as build‑pipeline execution or boilerplate generation—and share them securely across the organization. This promotes consistency, reduces repetitive configuration, and accelerates onboarding of new developers.
Auto Model Selection Generally Available in Copilot CLI
Copilot auto model selection became generally available in GitHub Copilot CLI on April 17. When set to “auto,” Copilot dynamically chooses the most efficient model (e.g., GPT‑5.4, GPT‑5.3‑Codex, Sonnet 4.6, Haiku 4.5) based on the user’s plan and policies. Paid subscribers receive a 10% discount on premium‑request multipliers when using the auto setting.
Why it matters for enterprise teams: Automating model selection removes decision overhead for developers and helps optimize cost‑performance trade‑offs. For organizations with mixed‑plan users, the feature ensures each developer gets the most appropriate model without manual tuning.
Visual Studio 2026 April Update Adds C++ Tools, Custom Agents, and Skill Auto‑Discovery
Visual Studio 2026 version 18.5.0, released on April 14, brings several Copilot‑related enhancements:
- C++ code editing tools for agent mode are now generally available by default, helping Copilot navigate C++ codebases more effectively.
- Agent skills auto‑discovery automatically finds and uses skills defined in the repository or user profile.
- Custom agents can be created and stored in the workspace or user directory, appearing in the agent picker with tailored instructions, tools, and model preferences.
- Customizable Copilot keyboard shortcuts let developers remap keys for accepting inline suggestions.
Why it matters for enterprise teams: Visual Studio remains a primary IDE for many enterprise .NET and C++ projects. These updates make Copilot more capable and customizable within that environment, reducing context‑switching and increasing productivity for Windows‑centric development teams.
Azure MCP Tools Built into Visual Studio 2022
Azure MCP tools are now built into Visual Studio 2022 as part of the Azure development workload—no separate extension required. Announced on April 15, this integration brings over 230 tools across 45 Azure services directly into GitHub Copilot Chat. Developers can manage Azure resources, deploy applications, query logs, and diagnose issues without leaving the IDE.
Why it matters for enterprise teams: Tight integration with Azure reduces the cognitive load of managing cloud infrastructure. Developers can stay focused on code while performing common operations via chat, speeding up DevOps workflows and reducing errors from manual command‑line steps.
Model Selection for Claude and Codex Agents on github.com
Model selection is now available for the Claude and Codex third‑party coding agents on github.com. As of April 14, users can choose among Anthropic models (Sonnet 4.6, Opus 4.6, Sonnet 4.5, Opus 4.5) for Claude and OpenAI models (GPT‑5.2‑Codex, GPT‑5.3‑Codex, GPT‑5.4) for Codex. This mirrors the flexibility already offered by Copilot Cloud Agent.
Why it matters for enterprise teams: Teams that use third‑party agents gain access to the latest model versions as they become available, allowing them to balance cost, speed, and capability according to each task’s requirements.
GitHub Platform Updates: OIDC, Rule Insights, Code Quality, and Security Scanning
Several platform improvements landed last week that affect Copilot‑related workflows:
- OIDC support for Dependabot and code scanning (April 14) eliminates the need to store long‑lived credentials as repository secrets.
- Rule insights dashboard and unified filter bar (April 16) provide visual analytics on rule evaluation activity.
- Code Quality improvements (April 14) add search by file path, bulk dismissal, and full diagnostic messages with Copilot Autofix suggestions.
- Security scanning updates include deployment context in alerts, linking code scanning alerts to GitHub Issues, and asynchronous SBOM exports.
Why it matters for enterprise teams: These enhancements streamline security and compliance processes, making it easier to manage policies, track issues, and automate fixes—areas where Copilot can already assist via Autofix and chat.
What to Watch
- Agent‑network filtering via group policy: VS Code 1.116 introduces group‑policy controls (
chat.agent.networkFilter) to restrict which domains agent tools can access. Enterprise administrators should evaluate this feature for tightening security around external tool calls. - SHA‑1 deprecation in HTTPS: GitHub will begin brownout testing on July 14, 2026, with full removal scheduled for September 15, 2026. Ensure all clients and integrations support modern TLS algorithms to avoid connectivity issues.
- Continued rollout of Claude Opus 4.7: The model will replace older Opus versions in the picker over the coming weeks. Monitor performance on complex tasks and adjust agent instructions if needed.
This week’s updates reflect GitHub’s dual focus on making Copilot more accessible (built‑in extension, auto model selection) and more sustainable (plan adjustments, granular enterprise controls). For teams investing in AI‑assisted development, staying abreast of these changes ensures they can maximize productivity while managing costs and security.
Check back next week for more developments across the Copilot ecosystem.