Summary
On July 29, 2026, GitHub moved Copilot code review support for agent skills and Model Context Protocol (MCP) servers to general availability for all Copilot Pro, Pro+, Business, and Enterprise users. Teams can add SKILL.md files under .github/skills to encode org-specific standards, and connect read-only MCP servers (issue trackers, docs, service catalogs) so reviews reflect internal context; comments now show when they were generated using skills or MCP.
What changed
Copilot code review's agent skills (SKILL.md in .github/skills) and MCP server connections graduated from public preview to general availability. All MCP tool calls during review are read-only. New attribution shows when a comment used skills or MCP context.
Why it matters
Lets organizations shape automated review with their own coding standards and internal systems instead of generic model judgment, making Copilot review more trustworthy in enterprise pipelines. Read-only MCP scoping and attribution address the governance and auditability concerns that block AI review adoption at scale.
Evidence excerpt
Copilot code review support for agent skills and Model Context Protocol (MCP) servers is now generally available for all Copilot Pro, Pro+, Business, and Enterprise users. All MCP tool calls performed by Copilot code review are limited to read-only.