GitHub Copilot app for Beginners: How to teach Copilot about your codebase
GitHub Copilot now supports the integration of custom instructions and repository-specific skills to refine AI responses. Users can now connect Model Context Protocol (MCP) servers, such as Playwright, to extend Copilot's functional capabilities.
Verified State Diff
Impact & Verification Analysis
Software developers, DevOps engineers, and enterprise teams utilizing GitHub Copilot.
This enables developers to reduce hallucination and improve relevance by grounding the AI in specific project requirements and allowing it to interact with external systems, significantly increasing the utility of Copilot in complex, non-standard codebases.
Full Fact Overview
This update marks a shift toward a modular, agentic architecture for GitHub Copilot. By enabling custom instructions, developers can inject project-specific context, coding standards, or architectural constraints directly into the LLM's prompt pipeline. The integration of the Model Context Protocol (MCP) allows Copilot to interface with external tools and data sources—specifically browser automation via Playwright—moving the tool beyond simple code completion into the realm of autonomous agentic workflows.