# Wan27 MCP [Health: Active]

**Category:** 🎥 Multimedia Process  
**Repository:** https://github.com/NotUsingGithub/wan27-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/wan27-mcp

## Description
Generate AI images and videos, browse models, estimate credits and track tasks with Wan27. MIT-licensed stdio client for Claude, Cursor and Codex. Requires an active paid subscription.

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

```json
"mcpServers": {
  "wan27-mcp": {
    "command": "npx",
    "args": ["-y","@wan27/cli"]
  }
}
```

## Documentation

## What Wan27 MCP MCP server does

Wan27 MCP MCP server gives an AI assistant access to image and video generation through a wan2-7.io account. It supports text-to-image, text-to-video, and image-to-video requests, provided that the selected model accepts the requested inputs. Generation uses the same account credits and prices as the website, so requests can consume the account’s existing balance.

The server also exposes account and model information before generation. An assistant can check account status, list available models, retrieve a model’s options and JSON Schema, estimate a request’s cost, and inspect previously created tasks. These capabilities make it possible to validate a request and review expected credit usage before starting a job.

## How it works

The package runs as a local stdio MCP server. After authentication, an MCP client launches the CLI with its `mcp` command and communicates with the process through standard input and output. The README documents setup commands for Claude and Codex, and also provides an npx-based JSON configuration for clients that accept that format.

The available MCP tools are `account_status`, `list_models`, `get_model`, `estimate_generation`, `generate_image`, `generate_video`, `list_tasks`, and `get_task`. Image-to-video workflows require an uploaded first frame URL. The CLI can upload a local file and return a URL for use in a later request.

Generation is asynchronous. A request produces a task that can be listed or inspected later. The CLI can wait for completion, while a wait timeout only stops the local wait and does not cancel the task. If a network error occurs, retrying with the original request ID returns the same task without charging it twice. Failed tasks are refunded automatically.

## Setup and configuration

Install Node 22 or newer on macOS, Linux, or WSL, then install the package globally with `npm install -g @wan27/cli`. Sign in using `wan27 auth login` before starting the MCP process. The account must have an active paid subscription; free accounts, unpaid trials, and credit-only purchases do not provide access.

Credentials are stored in the macOS Keychain by default. The CLI can instead use a private file store with `--credential-store file`. `WAN27_CONFIG_DIR` changes the configuration directory, and `WAN27_ACCESS_TOKEN` accepts a pre-issued token for automation. Logging out revokes the current connection, while changing the website password revokes every connection.

## Tools and capabilities

Wan27 MCP MCP server supports these operations:

- Check account status and available access.
- Browse models and inspect each model’s supported options.
- Estimate the credit cost of a proposed generation.
- Generate images or videos from structured requests.
- Upload an image for an image-to-video request.
- List tasks and retrieve an individual task’s status or result.

The CLI can download a completed result by task ID. Model-specific request fields should be checked with the model inspection command because options vary by model.

## Limitations and notes

The service is hosted by wan2-7.io, while this repository contains the public client distribution. Media generation requires an active subscription and spends website credits. The documented client setup is local stdio rather than a hosted MCP endpoint. Supported runtime environments are Node 22+, macOS, Linux, and WSL.

_Full upstream README: https://allmcps.com/mcp/wan27-mcp/readme_