Lotus Wisdom — Claude Code MCP server: contemplative reasoning
An MCP server that walks a model through step-by-step contemplative reasoning, mixing analysis with meditation pauses. Useful for problems that need several angles.
claude mcp add --transport http lotus-wisdom https://lotus-wisdom-mcp.linxule.workers.dev/mcpWhat Lotus Wisdom does
The server exposes a lotuswisdom tool that breaks thinking about a problem into steps. Each step carries a tag from a set of wisdom domains: entry, skillful means, non-dual recognition, meta-cognitive, process flow, meditation. The first call with tag='begin' returns the whole framework; after that the model moves through tags until the status reads WISDOM_READY. A second tool, lotuswisdom_summary, gives a recap of the path taken.
Two prompts ship with it: contemplate, which takes a question, and deep-inquiry, which takes a topic. Clients that support MCP Prompts surface them as slash commands or a prompt picker. Results arrive as a text block plus structuredContent validated against outputSchema. In clients with MCP ext-apps, every step renders as an interactive living trace that follows the host's light or dark theme.
There are two ways to connect: the remote HTTP endpoint https://lotus-wisdom-mcp.linxule.workers.dev/mcp, or a local stdio package run through npx. The remote worker keeps no session state, so continuity between steps travels in the previousJourney parameter. Run the local server with LOTUS_DEBUG=true and it prints trace lines to stderr. The tool is marked read-only and idempotent, though those are hints for the host rather than a promise of no side effects: local calls update the journey held in memory, and the hosted worker collects usage analytics. It does not touch your files or outside business data.
What it helps with
- work through a tangled problem from several angles
- make the model pause before it answers
- run a long inquiry across a topic step by step
- get a recap of how the reasoning went
- attach a contemplative framework to any MCP client
Example prompts for the agent
- Work through this problem with lotuswisdom, start with tag='begin'
- Run the contemplate prompt on: should we rewrite the payments module
- Start a deep-inquiry on migrating to a new framework
How to install Lotus Wisdom
1Platform CLI: remote server
claude mcp add --transport http lotus-wisdom https://lotus-wisdom-mcp.linxule.workers.dev/mcp2Platform CLI
claude mcp add lotus-wisdom -- npx -y lotus-wisdom-mcp1Platform CLI: remote server
codex mcp add lotus-wisdom --url https://lotus-wisdom-mcp.linxule.workers.dev/mcp2Platform CLI
codex mcp add lotus-wisdom -- npx -y lotus-wisdom-mcpConfig file
{
"mcpServers": {
"lotus-wisdom": {
"url": "https://lotus-wisdom-mcp.linxule.workers.dev/mcp"
}
}
}{
"mcpServers": {
"lotus-wisdom": {
"command": "npx",
"args": [
"-y",
"lotus-wisdom-mcp"
]
}
}
}Config file
{
"servers": {
"lotus-wisdom": {
"type": "http",
"url": "https://lotus-wisdom-mcp.linxule.workers.dev/mcp"
}
}
}{
"servers": {
"lotus-wisdom": {
"type": "stdio",
"command": "npx",
"args": [
"-y",
"lotus-wisdom-mcp"
]
}
}
}1Platform CLI: remote server
gemini mcp add -t http lotus-wisdom https://lotus-wisdom-mcp.linxule.workers.dev/mcp2Platform CLI
gemini mcp add lotus-wisdom npx -y lotus-wisdom-mcpStep by step, with commands for each platform: How to add an MCP server to Claude Code, Codex, Cursor, VS Code, Gemini CLI and more
Questions about Lotus Wisdom
How do I install Lotus Wisdom in Claude Code?
For the remote server: claude mcp add --transport http lotus-wisdom https://lotus-wisdom-mcp.linxule.workers.dev/mcp. For the local one: claude mcp add lotus-wisdom -- npx -y lotus-wisdom-mcp.
What does Lotus Wisdom do?
It guides a model through step-by-step contemplative reasoning based on the Lotus Sutra framework: tags, wisdom domains, meditation pauses, and a final answer once the status turns WISDOM_READY.
How is the remote server different from the local one?
The remote worker holds no session state, so path continuity is passed through previousJourney. The local stdio package keeps the journey in memory and can print trace lines to stderr when LOTUS_DEBUG=true.
