Coding Tools
GitHub Copilot Visual Studio September Update
GitHub Changeloghigh confidence
Why it matters
- →BYOM now mandates the new Agent mode, breaking backward compatibility with previous configurations.
- →NuGet vulnerability fixes are automated via MCP servers, reducing manual security remediation steps.
- →Git agent integration allows pull request exploration directly within the Copilot Chat interface.
GitHub has released a September update for GitHub Copilot in Visual Studio 2026, introducing deeper Git integration and automated security fixes. Key features include a Git agent that summarizes pull requests with navigable links to comments and files, and a new NuGet vulnerability fixer that uses MCP servers to patch packages directly from the Error List. The update also enforces a breaking change for Bring Your Own Model (BYOM), which now requires the new Agent mode built on the GitHub Copilot SDK. Previously configured Ollama models must be re-added, and support extends to Microsoft Foundry, OpenAI, Anthropic, and Ollama providers.
Read originalThe story around this
Earlier coverage that leads up to this article, and what followed. Lines connect each piece to the closest one after it, converging here.
How we got here
- 1GitHub Copilot July Update for Visual Studio
GitHub Changelog · July 30, 2026 · Same story
- 2GitHub Copilot August Update Enhances Control
GitHub Changelog · August 28, 2026 · Same story
More in Coding Tools
Coding Toolscoding
Claude Code v2.1.316: Subagents and Gateway Budgets
Anthropic is tightening control over Claude Code's autonomous behavior with new environment variables that let you disable 'ultrathink' or cap subagent effort levels. For teams, the introduction of pooled group budgets on the gateway allows shared monthly spending limits across members, a practical fix for enterprise cost management. The release also patches critical bugs where shell commands would hang or crash on exit, stabilizing the developer experience significantly.
Claude Code Releases
Coding Toolscoding
llama.cpp b12022 adds ROCm 10 and Snapdragon support
This release quietly cements llama.cpp as the universal inference runtime by finally bringing ROCm 10.0 to Linux and Windows alongside CUDA 13.4, effectively closing the hardware gap for AMD users who were previously left behind. The inclusion of native Snapdragon support on Linux marks a significant step toward running local LLMs on ARM-based mobile chips without emulation overhead. While KleidiAI on Apple Silicon is disabled in this build, the broader expansion to diverse accelerators ensures that developers aren't locked into NVIDIA ecosystems anymore.