Skip to main content
Glama
arupa-inc

taru-mcp

by arupa-inc
OverviewSchemaSourceRelated ServersDiscussionsAdmin

taru-mcp

MCP server for taru — connect Claude Code or Codex to your team's shared knowledge graph.

Zero dependencies. Pure Node.js. Works with any MCP client.

Quick Setup

macOS / Linux

curl -fsSL https://raw.githubusercontent.com/arupa-inc/taru-mcp/main/setup.sh | bash

Windows (PowerShell)

irm https://raw.githubusercontent.com/arupa-inc/taru-mcp/main/setup.ps1 | iex

The script will ask you:

  1. Project folder name — creates the workspace directory

  2. AI client — Claude Code, Codex, or both

  3. Workspace token — taru workspace token (xxv_...). Can be entered later if you don't have it yet.

It handles npm init, taru-mcp installation, agent file (CLAUDE.md / AGENTS.md) setup, and MCP registration automatically.

Get your workspace token from the taru web console: Settings > API Key.

Related MCP server: idea-basin-mcp

Manual Setup

If the setup script doesn't work, follow these steps:

1. Install

mkdir my-project && cd my-project
npm init -y
npm install taru-mcp

2. Copy agent instructions

# For Claude Code
cp node_modules/taru-mcp/samples/CLAUDE.md ./CLAUDE.md

# For Codex
cp node_modules/taru-mcp/samples/AGENTS.md ./AGENTS.md

3. Register MCP server

# Claude Code
claude mcp add taru -- node node_modules/taru-mcp/bin/taru-mcp.mjs --token xxv_your_token

# Codex
codex mcp add taru -- node node_modules/taru-mcp/bin/taru-mcp.mjs --token xxv_your_token

4. Or configure MCP manually

If claude mcp add doesn't work, create .mcp.json in your project root:

{
  "mcpServers": {
    "taru": {
      "command": "node node_modules/taru-mcp/bin/taru-mcp.mjs",
      "args": ["--token", "xxv_your_token"]
    }
  }
}

For Claude Code global config (~/.claude.json):

{
  "mcpServers": {
    "taru": {
      "command": "node node_modules/taru-mcp/bin/taru-mcp.mjs",
      "args": ["--token", "xxv_your_token"]
    }
  }
}

Tools

Once connected, these MCP tools are available to the AI:

Tool

Description

search_graph

Search the knowledge base by natural language query

read_full_document

Read full document content by UUID

store_document

Store a document or opinion with auto-conflict detection

list_documents

List all documents in the workspace

list_conflicts

View pending knowledge conflicts

rebalance

Merge similar keywords, clean up orphan nodes

How it works

taru-mcp is a thin proxy: it reads JSON-RPC (MCP protocol) from stdin, forwards each request to the taru server over HTTP, and writes the response to stdout. No database, no AI calls — just I/O.

MCP Client (Claude/Codex)
    ↕ stdin/stdout (JSON-RPC)
taru-mcp (this package)
    ↕ HTTP POST
Taru Server (knowledge graph + embeddings + graph DB)

License

MIT

This server cannot be deployed

One endpoint for every MCP toolRayrun runs every MCP server behind one URL and holds the credentials. Each call is decided per client, tool, and argument, then recorded.ray.runAd

Maintenance

ActivityInactive
ResponsivenessNo issues

Resources

Related MCP Connectors

Related MCP Servers

  • context-mcp

    vibhasdutta
    A
    license
    A
    quality
    B
    maintenance
    Provides persistent memory and a codebase knowledge graph for AI coding assistants, enabling shared context across multiple tools like Claude, Cursor, and ChatGPT, with significant token reduction.
    5
    36 npm
    MIT

  • MCP Memory Server

    beach55607-max
    A
    license
    Not graded
    quality
    D
    maintenance
    Provides long-term memory for AI coding agents, enabling them to remember, search, and organize information across sessions and platforms like Claude Code, ChatGPT, and Cursor.
    8 npm
    8
    MIT

  • Multimodal RAG MCP Server

    brett-hardiman
    A
    license
    Not graded
    quality
    D
    maintenance
    Connects Claude to a personal knowledge base with hybrid search over code, documents, and images. Allows the AI assistant to retrieve and answer from your own files during conversations.
    1
    MIT