Awesome MCP › Software Development

hopx-ai/mcp

⭐ 156 Python repository created 2025-11-10

HOPX MCP Server gives an MCP client the ability to run code in managed cloud sandboxes rather than on the user machine. Four languages are supported — Python, JavaScript, Bash and Go — and the Python environment ships with pandas, numpy and matplotlib preinstalled, which makes data analysis the intended primary use: a model writes an analysis script, runs it against uploaded data and reads back results including generated plots. Beyond one-shot execution the server supports file operations inside the container, arbitrary system commands, and background or long-running processes, so a session can span several steps against the same environment. Containers are destroyed automatically after use, which bounds both the security exposure and the cost of a forgotten sandbox. It is a Python project requiring Python 3.14 or newer and MCP 1.21 or newer, run with uvx hopx-mcp and configured with a HOPX_API_KEY environment variable obtained from hopx.ai; the README gives configuration blocks for Claude Desktop and other IDEs. It is MIT licensed and registered under the io.github.hopx-ai/hopx-mcp MCP name. Because execution happens in HOPX's cloud rather than locally, the trade-off is straightforward: no local Docker requirement and no local risk, in exchange for a hosted dependency and an API key. The audience is analysts and developers who want an agent that can actually execute and verify code.

https://github.com/hopx-ai/mcp

mcpsandboxcode-executionpythonclouddata-analysiscontainers