oatpp
Anthropic’s Model Context Protocol implementation for Oat++
claude mcp add --transport stdio oatpp-oatpp-mcp docker run -i oatpp/oatpp-mcp \ --env OATPP_MCP_PROMPTS="enable prompts as per MCP spec" \ --env OATPP_MCP_RESOURCES="allow access to resources" \ --env OATPP_MCP_TRANSPORT="STDIO or SSE (select via image/variant)"
How to use
oatpp-mcp is an MCP server implementation for the Oat++ framework. It enables Model Context Protocol features such as prompts, resources, and tools to be exposed for interaction with large language models. The server supports multiple transport modes, including STDIO and HTTP SSE, allowing you to feed prompts and resources to an LLM-based agent and receive structured responses according to the MCP spec. You can extend the server by adding prompts, resources, and tools in your Oat++ application, and then query the API using the generated MCP interfaces from tools like LLMs or downstream automation.
How to install
Prerequisites:
- A working C++ toolchain (compiler, CMake)
- The oatpp library/module installed as a dependency for your project
- Git for cloning the repository
Install steps:
-
Clone the repository: git clone <repository-url> cd <repository-folder>
-
Build the project (as per the repository’s guidance): mkdir build && cd build cmake .. make install
-
(Optional) Run tests located under test/ in the repository to verify setup: ctest --output-on-failure
-
Start the MCP server via your preferred transport (STDIO or SSE) once built and configured in your environment. The exact runtime invocation will depend on how you choose to expose the server (e.g., via a dedicated executable or integrated into your Oat++ HTTP server). See the Examples section in the README for sample usage patterns.
Additional notes
Notes:
- This MCP server is built around the Oat++ framework; ensure you have the oatpp dependencies properly installed before building.
- When running via STDIO, redirect logging if needed to avoid mixing logs with MCP traffic.
- The server can be extended by adding prompts, resources, and tools as shown in the Examples section of the README (e.g., adding CodeReview prompts, File resources, and Logger tools).
- If you intend to containerize, a Docker image or compose setup can simplify deployment; the README demonstrates a straightforward approach to building and wiring the server into an environment.
Related MCP Servers
dynamic-shell
Dynamic Shell Command MCP Server
claude-code
MCP Server connects with claude code local command.
docmole
Dig through any documentation with AI - MCP server for Claude, Cursor, and other AI assistants
mcp-jest
Automated testing for Model Context Protocol servers. Ship MCP Servers with confidence.
mcpx
MCP client with connection pooling
mcp-core
core utilities and tools for model context protocol