Yarrow
An I-Ching oracle MCP server for casting hexagrams, looking up bilingual classical sources, and generating grounded Wilhelm/Baynes-style reflections.
README
Yarrow — An I-Ching oracle for the agent era
Yarrow lets you (and your agents) consult the Book of Changes — cast a hexagram, read the classical sources, and get a Wilhelm/Baynes-style reflection. Offline-first iOS app, plus a public MCP server so any LLM agent can reach the same oracle.
"The I-Ching does not predict; it mirrors. The question you bring decides the reading."
What's in this repo
| Path | What |
|---|---|
src/ |
Cloudflare Workers source — public MCP endpoint + iOS app proxy |
src/data_hexagrams.json |
64 hexagrams with bilingual classical sources (杨天才 / 朱熹 / 断易天机 / 白话题解) |
src/hexagram.ts |
Deterministic casting (coins / yarrow stalks / random), seeded by FNV-1a + mulberry32 |
src/mcp.ts |
MCP JSON-RPC 2.0 handler (3 tools, 30 readings/day per anonymous caller) |
.claude-plugin/marketplace.json |
Claude Code / Codex plugin marketplace manifest |
skills/yarrow/SKILL.md |
The Yarrow skill — pluggable into any Claude Code or Codex install |
npm/yarrow-mcp/ |
@shaozhengkun/yarrow-mcp — stdio bridge for clients that don't speak HTTP MCP |
submissions/ |
Drafts for modelcontextprotocol/servers, Smithery, mcp.run |
INTEGRATIONS.md |
Copy-paste configs for Claude Code / Claude Desktop / Codex / Hermes |
The iOS source lives separately at yarrow-ios (TODO).
For users
📱 iOS app
Download on the App Store → · $4.99 · iOS 17+ · no tracking, no account.
All 64 hexagrams and their classical commentary ship in the app; AI reading is optional and runs through this Worker.
🤖 Agents
Claude Code — install the plugin:
claude plugin marketplace add https://github.com/shaozhengkun123/yarrow
claude plugin install yarrow
Claude Desktop — add to ~/Library/Application Support/Claude/claude_desktop_config.json:
{
"mcpServers": {
"yarrow": { "url": "https://yarrow-oracle.shaozhengkun.workers.dev/mcp" }
}
}
Codex CLI / Hermes / older Claude Desktop — use the stdio bridge:
npx @shaozhengkun/yarrow-mcp
Or pin it in your client's config — see INTEGRATIONS.md for every platform.
Anyone with curl —
curl -s -X POST https://yarrow-oracle.shaozhengkun.workers.dev/mcp \
-H 'Content-Type: application/json' \
-d '{"jsonrpc":"2.0","id":1,"method":"tools/call","params":{
"name":"divine",
"arguments":{"question":"Should I take the new job?","locale":"en"}
}}'
The MCP tools
| Tool | Purpose |
|---|---|
cast_hexagram |
Toss the coins / cast the stalks. Returns primary hexagram + moving lines + changing hexagram, all metadata. Deterministic if you pass seed. |
lookup_hexagram |
Get a hexagram's classical text — Wilhelm-style header, judgment, image, and 杨天才 / 朱熹 / 断易天机 / 白话 commentaries. |
divine |
One-shot reading: casts, looks up, and generates a Wilhelm/Baynes-style reflection grounded in the sources. 30/day per anonymous caller. |
Self-hosting
This Worker is MIT — fork, deploy your own, point your agents at your URL.
git clone https://github.com/shaozhengkun123/yarrow
cd yarrow
npm install
npx wrangler login
npx wrangler kv:namespace create QUOTA # copy id into wrangler.toml
echo -n '<your OpenAI-compatible chat URL>' | npx wrangler secret put PROXY_URL
echo -n '<your API key>' | npx wrangler secret put PROXY_KEY
npx wrangler deploy
Your endpoint will be https://yarrow-oracle.<your-subdomain>.workers.dev.
Boundaries
Yarrow is for reflection, not prediction. The system prompt baked into the Worker refuses:
- fortune-telling / specific future predictions
- medical, legal, or financial advice
- requests to "reveal what will happen"
The classical voice is direct; the framing is Jungian. This is the line we draw, in the iOS app and the MCP both.
License
MIT for code; classical sources are in the public domain. See LICENSE.
Recommended Servers
playwright-mcp
A Model Context Protocol server that enables LLMs to interact with web pages through structured accessibility snapshots without requiring vision models or screenshots.
Audiense Insights MCP Server
Enables interaction with Audiense Insights accounts via the Model Context Protocol, facilitating the extraction and analysis of marketing insights and audience data including demographics, behavior, and influencer engagement.
Magic Component Platform (MCP)
An AI-powered tool that generates modern UI components from natural language descriptions, integrating with popular IDEs to streamline UI development workflow.
VeyraX MCP
Single MCP tool to connect all your favorite tools: Gmail, Calendar and 40 more.
graphlit-mcp-server
The Model Context Protocol (MCP) Server enables integration between MCP clients and the Graphlit service. Ingest anything from Slack to Gmail to podcast feeds, in addition to web crawling, into a Graphlit project - and then retrieve relevant contents from the MCP client.
Kagi MCP Server
An MCP server that integrates Kagi search capabilities with Claude AI, enabling Claude to perform real-time web searches when answering questions that require up-to-date information.
E2B
Using MCP to run code via e2b.
Neon Database
MCP server for interacting with Neon Management API and databases
Exa Search
A Model Context Protocol (MCP) server lets AI assistants like Claude use the Exa AI Search API for web searches. This setup allows AI models to get real-time web information in a safe and controlled way.
Qdrant Server
This repository is an example of how to create a MCP server for Qdrant, a vector search engine.