Wenxiang MCP
Bring the cultural interpretations of the 64 hexagrams into AI, offering fresh perspectives for observation and reflection.
Connect to your AI
Choose a client, copy the configuration, then check that the tools have loaded.
Public MCP endpoint
https://wenxiang-mcp.328558608.workers.dev/mcp
Use this endpoint in an MCP client that supports Streamable HTTP. The public service needs no login or API key. Check connection status and tool availability in your client.
Once the public endpoint is published, this page will provide its address and client configurations. You can also deploy your own service from the source package below.
MCP connects tools to your AI client. Add the public endpoint, then confirm that your client discovers the tools.
Choose your client and copy the configuration
codex mcp add wenxiang --url https://wenxiang-mcp.328558608.workers.dev/mcp
codex mcp listRun the commands in your terminal, then start or reopen a Codex session. Use /mcp to see the loaded tools. If a configuration with this name already exists, check its endpoint first. OpenAI documentation ↗
claude mcp add --transport http wenxiang --scope user https://wenxiang-mcp.328558608.workers.dev/mcp
claude mcp get wenxiang--scope user saves the connection to your user configuration. Reopen Claude Code and use /mcp to check connection status and tools. These commands are for the Claude Code terminal client. Claude Code documentation ↗
{
"mcpServers": {
"wenxiang": {
"url": "https://wenxiang-mcp.328558608.workers.dev/mcp"
}
}
}Merge this into your user ~/.cursor/mcp.json or project .cursor/mcp.json, keeping your existing server configurations. Save, reload as Cursor prompts, and check the tools in MCP settings. Cursor documentation ↗
Run your own service
Start with the source package and its locked dependencies, then publish Wenxiang MCP to your Cloudflare Worker.
View deployment stepsTools and responses
Three simple cultural tools return hexagram details, prewritten interpretations and prompts for reflection. How you interpret them is up to you.
| Tool | What it does | Inputs |
|---|---|---|
draw_hexagram | A chance encounterDraw a random hexagram and view its name and six lines, offering a cultural lens for observing a question. | No parameters |
get_hexagram | Explore a hexagramRead a hexagram by number to explore its structure and cultural meaning. | hexagram_id: 1–64 |
explain_hexagram | A new perspective for reflectionRead prewritten interpretations and reflective prompts, then return to your own understanding. | hexagram_id: 1–64 |
A cultural experience for reflection. The tools do not receive your personal question or use a model to infer outcomes. Random hexagrams and cultural interpretations inspire observation and reflection, with no promise of predictions or decision outcomes. The interface and cultural interpretations of all 64 hexagrams support 简体中文, 繁體中文 and English. All three MCP tools accept a language parameter; classical Chinese source text is kept separately.
Language selection in tool arguments:language accepts zh-CN / zh-Hant / en.
Try a prompt
Example prompt
Set language="en" when calling the tools. Draw a random hexagram and show its name and six lines. Based on its cultural interpretation, give me three open-ended questions for reflection, without predicting outcomes.
Cultural readingLook up hexagram 1 and explain its six-line structure and the cultural meaning of Qian.
Another perspectiveRead a hexagram’s prewritten interpretation and turn its reflective ideas into journal prompts.
Deploy your own service
Publish Wenxiang MCP to your Cloudflare Worker. The service includes cultural texts for all 64 hexagrams and read-only tools. After deployment, connect your client through its HTTPS endpoint.
The service returns prewritten content. It needs no database or model API key.
Download and extract the source
Download and extract the source package locally, then enter the
Download Wenxiang MCP sourcewenxiang-mcpdirectory. Its README covers versions, configuration and verification.Install dependencies and publish to Cloudflare
Run these commands in the source directory.
npm ciuses the locked versions in the package. Once tests and build checks pass, follow Wrangler’s prompts to authorize Cloudflare access. Before publishing, set your service’s Worker name inwrangler.tomland check the rate-limit namespace.npm ci npm test npm run check npx wrangler login npm run deployUse your own endpoint
Take the actual HTTPS service address from the deployment output and append
/mcp. In the source directory, runnpm run smoke --followed by this MCP endpoint to complete the SDK check. Then return to “Connect to the public service” and replace the example URL with your endpoint.After connecting, confirm discovery ofdraw_hexagram,get_hexagramandexplain_hexagram, then try reading a hexagram.
This package deploys the MCP tool service. For a custom domain, configure SITE_ORIGIN as described in the README. Browser connections across origins require an explicit allowed origin. Check quotas and the public endpoint against your Cloudflare configuration.