Advances in the field of artificial intelligence have led to the introduction of autonomous applications (AI Agents), capable of interacting with external application interfaces. While these AI agents have seen increased adoption, a recurring challenge is a standard communication mechanism between these agents and external applications. The Model Context Protocol (MCP) was introduced as a standardized interface to address this challenge. However, little is known about practical issues experienced by developers in its adoption. Our goal for this study is to identify the challenges faced by developers in the adoption of MCP and the effort expended in addressing them. To achieve this, we analyze bug reports and pull requests on MCP reference servers. Our findings reveal that issues relating to file system operations (21.32%) are the most prevalent, followed by data validation and type issues (13.97%). Also, we find that developers spend more effort addressing connection and communication issues. Our work presents the MCP issues developers encounter and provides insight into the extent of work required to resolve them. Our findings can provide MCP repository maintainers and contributors with a structured view of developer issues and help them make informed prioritizations.