Artificial intelligence can greatly speed up development, but is limited by its training data and public sources. Developers often need access to internal systems like project trackers, databases, and documents. The Model Context Protocol (MCP) provides AI with secure access to internal tools, allowing for comprehensive assistance within developer workflows. MCP is an open standard that connects AI with data and tools, acting as a secure two-way channel. MCP servers act as bridges to databases, APIs, cloud services, and other external tools, enabling AI tools to access real-time information and perform actions. MCP simplifies AI integration with systems by providing a standardized interface, unlike individual API integrations. GitLab Duo Agent Platform now supports MCP, allowing developers to access their entire toolkit through natural language in their IDE. GitLab functions as both an MCP client, accessing external tools, and an MCP server, allowing AI tools to connect to GitLab instances. Setting up GitLab MCP client involves enabling it in group settings and configuring MCP servers in a JSON file. An example project demonstrates MCP functionality in a simulated enterprise setup for incident response. GitLab also offers MCP server capabilities, enabling AI tools to securely connect to GitLab instances and work with development data. AI tools can then fetch and act on GitLab data like issues and merge requests using natural language. This bidirectional MCP support enhances developer workflows and ensures AI provides meaningful assistance.