鸣潮

You can run WuWa MCP Server to fetch character and artifact information from the WuWa game and receive results formatted in Markdown. It’s optimized for use with large language models and supports both local STDIO and remote streamable HTTP transports for flexible deployment.

How to use

Start by running the server locally in STDIO mode for interactive usage with clients like Claude Desktop, or switch to an HTTP transport when deploying to the cloud. You will interact with the MCP server through your client, which will request character, artifact, or profile data and receive Markdown-formatted responses suitable for LLM consumption.

How to install

Prerequisites: you need Python 3.12+ installed on your system. You also need a way to launch the MCP server locally (the provided runtime uses the uv utility). You may optionally use Smithery to install the server remotely.

Install via Smithery (remote installation):

npx -y @smithery/cli@latest install @jacksmith3888/wuwa-mcp-server --client claude --key YOUR_SMITHERY_KEYs

Run locally

To run in STDIO mode (local interactive mode):

uv run python -m wuwa_mcp_server.server

Run with HTTP transport

To run with a streamable HTTP transport (cloud-friendly):

Note: HTTP mode requires the environment to enable HTTP streaming. The following command shows the pattern used when starting in HTTP mode through the same runtime.

TRANSPORT=http uv run python -m wuwa_mcp_server.server

Available tools

get_character_info

Fetches detailed information about a specified WuWa character and returns it as Markdown for easy consumption by language models.

get_artifact_info

Fetches detailed information about a specified artifact set and returns it as Markdown.

get_character_profile

Fetches a character's profile information and returns it as Markdown.