GitHub Copilot Adds Enterprise-Managed Permissions for Agent Operations

GitHub shipped enterprise-managed permissions for GitHub Copilot’s agentic features on September 9, 2026, giving administrators central control over which agent operations are automatically blocked, require human approval, or can proceed without a prompt.

Details

• What’s new: enterprise admins can now set org-wide policies covering shell commands, file reads and edits, and network domain access for Copilot’s agent operations
• Enforcement that sticks: enterprise-level restrictions cannot be overridden by individual user settings, workspace configurations, or any auto-approval mechanism a developer has turned on locally
• Team-specific policies: enterprises can build multiple, differentiated permission profiles for different teams rather than a single blanket rule
• Where it applies: the managed permissions are enforced across the GitHub Copilot app, the GitHub Copilot CLI, and Visual Studio Code sessions that use Agent Host
• Goal: let security-conscious organizations keep fine-grained guardrails on sensitive agent actions without shutting down agentic workflows entirely

What happened next

The feature is generally available now, with configuration documented under GitHub’s enterprise-managed settings. It follows a string of admin-control-focused Copilot changelog entries through early September — including default-model policies, content-exclusion enforcement, and budget-expiration controls — that together point to GitHub prioritizing enterprise governance tooling as Copilot’s agent capabilities keep expanding.