GitHub Copilot Code Review Agent Skills and MCP Servers Now Generally Available
Key Takeaways
- ▸Agent Skills and MCP server support for Copilot code review reaches general availability across all subscription tiers
- ▸Developers can now create custom SKILL.md files to embed team-specific tools, coding standards, and repository context into code reviews
- ▸MCP servers enable seamless integration with third-party platforms (issue trackers, docs, service catalogs) while enforcing read-only access for security
Summary
GitHub Copilot's code review feature now generally supports Agent Skills and MCP (Model Context Protocol) server connections, enabling developers to bring custom team tools, coding standards, and external context directly into code reviews. Previously available in public preview, these capabilities are now rolling out to all Copilot Pro, Pro+, Business, and Enterprise users.
Agent Skills allow teams to create custom SKILL.md files in their repositories that extend Copilot's analysis with organization-specific tools and standards. Meanwhile, MCP server connections integrate third-party platforms—such as issue trackers, documentation systems, and service catalogs—directly into the review process, with all tool calls limited to read-only access for security. Attribution has been added to indicate when Copilot generates comments using agent skills or MCP context, providing transparency into the analysis sources.
- GitHub and Playwright MCP servers will be enabled by default; existing configurations from the public preview automatically carry over
- Code review comments now attribute which insights came from agent skills or MCP servers, improving transparency and debugging
