In the process to test our auto-generated MCP tools, we discovered critical limitations. This article shares practical insights from running MCP tools with real AI Agents.

In Parts 1 and 2, we built a unified generation pipeline where one proto file produced gRPC services, REST endpoints, OpenAPI specs, and an MCP server. Everything worked technically, the tools existed and could be called. But when we began testing with AI Agents, we discovered a critical bottleneck: the AI frequently failed to choose the right tool or collect the required parameters correctly.

In Part 1, we demonstrated how gRPC transcoding can unify gRPC and REST API definitions. This article takes the next step, showing how to generate an MCP server and its tools from the same Protobuf file that defines those API endpoints.

This article explains how to generate REST APIs from Protobuf using gRPC transcoding. It introduces the basic process and tools needed to expose gRPC services as RESTful endpoints. This is the foundation for building an MCP server, which will be covered in the next part.