# pythonanywhere/pythonanywhere-mcp-server [Health: Active]

**Category:** ☁️ Cloud Platforms  
**Repository:** https://github.com/pythonanywhere/pythonanywhere-mcp-server  
**GitHub Stars:** 17  
**Views:** 3  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/pythonanywhere-pythonanywhere-mcp-server

## Description
MCP server implementation for PythonAnywhere cloud platform.

## Claude Desktop Quick Installation
Install path detected from listing signals. Uses `uvx` (confidence: high):

```json
"mcpServers": {
  "pythonanywhere-mcp-server": {
    "command": "uvx",
    "args": ["pythonanywhere-mcp-server"],
    "env": {
      "API_TOKEN": "",
      "PYTHONANYWHERE_USERNAME": ""
    }
  }
}
```

**Requires environment variables:** `API_TOKEN`, `PYTHONANYWHERE_USERNAME` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What pythonanywhere/pythonanywhere-mcp-server MCP server does

The pythonanywhere/pythonanywhere-mcp-server MCP server gives MCP-compatible AI clients access to selected PythonAnywhere account operations. It works with the PythonAnywhere API through the `pythonanywhere-core` package and presents those operations as tools that a client can call.

Supported areas include:

- Reading, uploading, deleting, and listing files and directory trees.
- Inspecting log files through the same file access used for other files.
- Creating, deleting, reloading, and listing ASGI web applications.
- Creating, deleting, reloading, patching, listing, and inspecting WSGI web applications.
- Listing, creating, updating, and deleting scheduled tasks.

This makes the pythonanywhere/pythonanywhere-mcp-server MCP server relevant for workflows that combine code or log inspection with PythonAnywhere deployment administration. It is also suitable for clients that need to update scheduled tasks without building a separate PythonAnywhere API integration.

## How it works

The server uses the Python MCP SDK and the `pythonanywhere-core` library, which wraps part of the PythonAnywhere API. An MCP client starts the server as a local stdio process, then sends tool requests over the MCP connection. PythonAnywhere authenticates those requests with an API token associated with the configured account and username.

The implementation covers a subset of the PythonAnywhere API rather than every available platform operation. Its supported surface may expand as the underlying project develops.

## Setup and configuration

`uv` must be installed and available on the system `PATH`. The standard command is:

```bash
uvx pythonanywhere-mcp-server
```

Configure these environment variables in the MCP client:

- `API_TOKEN`: the PythonAnywhere API token.
- `PYTHONANYWHERE_USERNAME`: the PythonAnywhere account username.
- `PYTHONANYWHERE_SITE`: optionally set to `eu.pythonanywhere.com` for accounts hosted on the EU site; the default is `www.pythonanywhere.com`.

The README documents configurations for Claude Desktop, Claude Code, Cursor, and GitHub Copilot in PyCharm. Claude Desktop can also load the project’s downloadable MCP Bundle file. For manual stdio configurations, use `uvx` as the command and `pythonanywhere-mcp-server` as its argument.

## Tools and capabilities

The pythonanywhere/pythonanywhere-mcp-server MCP server can expose file operations useful for editing application files or reviewing logs. Its web application tools distinguish between ASGI and WSGI deployments and provide different operation sets for each type. WSGI support includes retrieving application information and applying patches, while ASGI support includes application creation and reload operations.

Scheduled-task tools support the full basic lifecycle: discovery, creation, modification, and removal. Because a scheduled task can execute an arbitrary command if configured to run soon after creation, task changes require particular care and human review.

## Limitations and notes

Credentials provide access to the configured PythonAnywhere account, so file deletion, application changes, and task modifications should not be delegated without appropriate oversight. The project specifically warns about direct LLM access to account resources and recommends human supervision for sensitive operations.

The server does not represent the complete PythonAnywhere API. It currently relies on the subset implemented by `pythonanywhere-core`. When multiple MCP servers run together, review their combined access, especially when another server can retrieve untrusted external content.

The README mentions using a separate time MCP server to help models handle scheduled-task timing, but that is not part of this project and is not required to start the pythonanywhere/pythonanywhere-mcp-server MCP server.

_Full upstream README: https://allmcps.com/mcp/pythonanywhere-pythonanywhere-mcp-server/readme_