nathandelisle/mcp-codeexec-server
If you are the rightful owner of mcp-codeexec-server and would like to certify it and/or have it hosted online, please leave a comment on the right or send an email to dayong@mcphub.com.
The MCP Code-execution Server is a secure environment for executing Python and Node.js code snippets within a Docker container, ensuring isolation and resource control.
MCP Code-execution Server
MCP server which exposes a single tool (execute_code) to run Python or Node snippets in a hardened Docker container. Integrates over MCP stdio by default but HTTP is available if you need it.
Need Python 3.11+ and Docker desktop to run.
To run: python -m venv .venv source .venv/bin/activate # on Windows: ..venv\Scripts\activate pip install -r requirements.txt ./scripts/build.sh # or, scripts\build.bat python tests/smoke_client.py # test MCP handshake, execute_code for Python and Node, file IO and stdin piping
To run server for a client, use python server.py stdio or, if you prefer an endpoint, python server.py http --host 127.0.0.1 --port 3333
The tool takes an argument object called request. Specify language as python or node, provide code as a string, and optionally include stdin, a list of auxiliary files (each with path and content), and some resource knobs (e.g. timeout_seconds)
Executions run offline by default, in a container with read-only root filesys, no Linux, and cgroup limits for CPU, memory, process count. In practice this means code cannot reach the network unless you explicitly allow it; it cannot modify system files; it cannot perform privileged operations; and it cannot exhaust the host.
On Windows, stdin is passed via a file inside the workspace so Docker named pipes behave consistently