Metadata-Version: 2.4
Name: envhub-mcp
Version: 0.1.0
Summary: MCP server providing containerized conda environments via Podman
Project-URL: Homepage, https://github.com/shenny/envhub
Project-URL: Repository, https://github.com/shenny/envhub
Author-email: shenny <406205391@qq.com>
License-Expression: MIT
Keywords: ai-agent,conda,container,mcp,podman
Classifier: Development Status :: 3 - Alpha
Classifier: License :: OSI Approved :: MIT License
Classifier: Programming Language :: Python :: 3
Classifier: Topic :: Scientific/Engineering
Classifier: Topic :: Software Development :: Libraries
Requires-Python: >=3.10
Requires-Dist: mcp>=1.9.0
Requires-Dist: pyyaml>=6.0
Description-Content-Type: text/markdown

# envhub

MCP server providing containerized conda environments via Podman for AI agents.

## Prerequisites

- Python >= 3.10
- [Podman](https://podman.io/docs/installation)

## Install

```bash
pip install envhub-mcp
```

## Build Container Image

```bash
# Find the docker directory
python -c "import envhub; print(envhub.__file__)"
# Build (from the package's docker/ directory)
podman build -t envhub:latest <path-to-envhub>/docker/
```

## Usage

```bash
envhub --workdir /path/to/work
```

Multiple working directories:

```bash
envhub --workdir /path/a --workdir /path/b
```

## MCP Tools

| Tool | Description |
|------|-------------|
| `run(command, env, timeout)` | Execute shell commands in a conda environment |
| `list_envs()` | List available conda environments |
| `list_workdirs()` | List mounted working directories |

## Environments

- **default** — Data analysis + visualization (pandas, scikit-learn, matplotlib, plotly, R/ggplot2)
- **plot** — Professional plotting (matplotlib, seaborn, plotly, plotnine, R/ggplot2, cairo)

## License

MIT
