# Wan27 MCP — AI Image and Video Generation [Health: Active]

**Category:** 💻 Developer Tools  
**Repository:** https://github.com/DieselNiu/wan27-mcp  
**GitHub Stars:** 0  
**Views:** 0  
**Installs:** 0  
**Upvotes:** 0  
**Directory Page:** https://allmcps.com/mcp/wan27-mcp-ai-image-and-video-generation

## Description
Generate AI images and videos, estimate credits and track tasks. Paid Wan27 subscription required.

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

```json
"mcpServers": {
  "wan27-mcp-ai-image-and-video-generation": {
    "command": "npx",
    "args": ["-y","@wan27/cli"],
    "env": {
      "WAN27_CONFIG_DIR": "",
      "WAN27_ACCESS_TOKEN": ""
    }
  }
}
```

**Requires environment variables:** `WAN27_CONFIG_DIR`, `WAN27_ACCESS_TOKEN` — the values above are empty placeholders; fill in real credentials before running (see the repository for what each one is for).

## Documentation

## What Wan27 MCP — AI Image and Video Generation does

Wan27 MCP — AI Image and Video Generation exposes the wan2-7.io generation service to MCP-compatible assistants. It supports image generation, video generation, model inspection, credit estimation, and task management through a local command launched by the Wan27 CLI.

The service uses the same account and credit balance as the wan2-7.io website. Generated images and videos therefore consume website credits at the account’s applicable prices. Before sending a generation request, an assistant can inspect account status and estimate the expected cost.

Supported workflows include text-to-image, text-to-video, and image-to-video generation. Image-to-video requests use an uploaded first frame supplied as a URL. The CLI also provides commands for downloading completed results.

## How it works

Install the CLI with npm, authenticate with `wan27 auth login`, and register the stdio server with an MCP client. The client starts `wan27 mcp` locally, while the CLI uses the authenticated wan2-7.io connection to handle requests.

The available MCP tools are:

- `account_status` for account and access information
- `list_models` and `get_model` for discovering models and their options
- `estimate_generation` for checking an estimated credit cost
- `generate_image` and `generate_video` for submitting generation requests
- `list_tasks` and `get_task` for checking submitted work

Generation requests return a task that can be checked later. A request ID can be reused after a network error so the same task is returned without a second charge. A timeout while waiting does not cancel the task, and failed tasks are refunded automatically.

## Setup and configuration

The documented runtime requirements are Node 22 or newer and macOS or Linux, including WSL. Install the package globally with `npm install -g @wan27/cli`, sign in once, then add the server with either `codex mcp add wan27 -- wan27 mcp` or `claude mcp add wan27 -- wan27 mcp`.

Clients that accept JSON MCP configuration can launch the server with `npx -y @wan27/cli mcp`. Authentication credentials are stored in the macOS Keychain by default. The `--credential-store file` option uses a private file with a 0700 directory and 0600 file permission model.

`WAN27_CONFIG_DIR` changes the configuration directory, while `WAN27_ACCESS_TOKEN` supplies a pre-issued token for automation. Logging out revokes the current connection. Changing the wan2-7.io website password revokes every connection.

## Tools and capabilities

Wan27 MCP — AI Image and Video Generation can discover a model’s complete options and JSON Schema through the CLI’s model inspection command. Requests are supplied as JSON, including fields such as `modelId`, `prompt`, duration, resolution, and an image URL where applicable.

The CLI examples identify `wan27-text-to-video`, `wan2.7-image`, and `wan27-image-to-video` model IDs. A separate upload command accepts a local frame and returns a URL that can be passed to an image-to-video request. Completed media can be retrieved with the task ID and an output path.

## Limitations and notes

Access requires an active paid Wan27 subscription. Credit packs by themselves, free accounts, and unpaid trials are not sufficient. A subscription with auto-renewal disabled remains usable until its paid period ends.

The MCP server is local and uses stdio; the provided material does not describe a hosted MCP endpoint. Image and video generation spend the account’s website credits, so assistants should check account status and estimate costs before generating. The repository is licensed under MIT and mirrors the published `@wan27/cli` executable distribution.

_Full upstream README: https://allmcps.com/mcp/wan27-mcp-ai-image-and-video-generation/readme_