GitHub Copilot cloud agent gets reasoning control and comment triggers
Delegate to the agent straight from a PR comment and dial reasoning per task — while six models, including Claude Opus 4.6, exit Copilot on Sept 1.
Copy markdown
Comment on a PR, and the agent takes it from there
Copilot's cloud agent can now be fired straight from an issue or PR comment: define the trigger text once under a repo's Agents → Automations tab, then a matching comment auto-generates docs, digs through a stack trace, or files follow-up refactor issues. Live on Pro, Pro+, and Max, and on Business/Enterprise once an admin turns on the cloud-agent policy.
Dial reasoning up or down per task
When you hand the cloud agent a job, you now pick a reasoning level right next to the model. Crank it up on a thorny bug and it thinks harder; keep it low on routine work to save tokens and credits. It's rolling out across every paid plan with cloud-agent access.
Six models leave Copilot on Sept 1 — check your defaults
GitHub is retiring Claude Opus 4.5 and 4.6, Sonnet 4.5 and 4.6, Gemini 3.1 Pro, and Raptor Mini across all Copilot surfaces on September 1. Repoint your workflows to Opus 4.7/4.8/5, Sonnet 5, Gemini 3.6 Flash, or MAI-Code-1-Flash before then. Annual individual subscribers keep Sonnet 4.6.
AI now writes your coverage workflow
A repo's Code Quality settings can now auto-generate a code-coverage workflow for you — no hand-written CI YAML. It's a one-click path to seeing which lines your tests actually exercise on each pull request.