Copilot code review: Agent skills and MCP now generally available

GitHub has moved Copilot code review support for agent skills and MCP servers from public preview into general availability. The feature is available to all Copilot Pro, Pro+, Business, and Enterprise users, as announced on the GitHub Changelog and The GitHub Blog.

In practical terms, Copilot code review can now use agent skills and MCP servers as part of how it inspects pull requests. Agent skills define specialized review behavior; MCP servers connect that review flow to external tools and context. The change is a product-scope expansion of code review, not a separate product line.

For engineers and builders already on Copilot Pro through Enterprise, review automation can lean on custom skills and MCP-backed context instead of a fixed, one-size review path. Teams that trialled the public preview can treat the capability as production-ready rather than experimental.

The GA release puts agent-skills-and-MCP-aware code review into the full paid Copilot tiers at once. That matters in a market where AI review is judged by how well it fits existing tooling and org-specific checks, not only generic style nits.

What to watch next is how teams wire real agent skills and MCP servers into day-to-day PR review on those tiers, and whether GitHub extends the same review surface beyond the currently listed Copilot plans. Until then, the actionable step is to enable and test skills plus MCP integrations under Copilot code review if you already hold Pro, Pro+, Business, or Enterprise.