@infopercept/invinsense-rag-mcp

@infopercept/invinsense-rag-mcp

MCP (Model Context Protocol) server that provides Claude and other AI assistants with semantic search access to InvinSense documentation using Vertex AI RAG Engine.

Installation

npm install -g @infopercept/invinsense-rag-mcp

Or use directly with npx (recommended for MCP):

npx @infopercept/invinsense-rag-mcp

MCP Configuration

Claude Code CLI

Add to ~/.claude/settings.json:

{
  "mcpServers": {
    "invinsense-rag": {
      "command": "npx",
      "args": ["@infopercept/invinsense-rag-mcp"],
      "env": {
        "RAG_API_URL": "https://your-rag-endpoint",
        "RAG_API_KEY": "your-api-key"
      }
    }
  }
}

Claude Desktop

Add to ~/.config/claude/mcp.json:

{
  "mcpServers": {
    "invinsense-rag": {
      "command": "npx",
      "args": ["@infopercept/invinsense-rag-mcp"],
      "env": {
        "RAG_API_URL": "https://your-rag-endpoint",
        "RAG_API_KEY": "your-api-key"
      }
    }
  }
}

Environment Variables

Variable	Description	Default
RAG_API_URL	RAG service endpoint	http://localhost:8080
RAG_API_KEY	API key for RAG service	(none)

Available Tools

rag_query

Semantic search through InvinSense documentation.

Search for "How does identity broker authentication work?"

rag_generate

Get AI-generated answers with citations using Gemini + RAG.

What authentication protocols does the Identity Broker support?

rag_corpus_info

Get RAG corpus information (document count, status).

rag_health

Check RAG service health and connectivity.

Architecture

┌─────────────────┐     ┌─────────────────┐     ┌─────────────────┐
│  Claude/MCP     │────▶│   MCP Server    │────▶│   RAG Service   │
│   Client        │     │  (this package) │     │  (your endpoint)│
└─────────────────┘     └─────────────────┘     └────────┬────────┘
                                                         │
                              ┌───────────────────────────┼───────────────────────────┐
                              │                           │                           │
                              ▼                           ▼                           ▼
                     ┌─────────────────┐        ┌─────────────────┐        ┌─────────────────┐
                     │  Vertex AI RAG  │        │   Google AI     │        │      GCS        │
                     │    Engine       │        │  Gemini 2.0     │        │   (Documents)   │
                     └─────────────────┘        └─────────────────┘        └─────────────────┘

RAG Service Setup

This MCP server connects to a RAG service backend. See the full repository for RAG service setup:

Prerequisites

• Node.js >= 20.0.0
• Google Cloud account with billing enabled
• Service account with Vertex AI permissions

Quick Start

# Clone the repository
git clone https://gitlab.invinsense.dev/infopercept/docs/pg-rag.git
cd pg-rag

# Install dependencies
npm install

# Configure environment
cp .env.example .env
# Edit .env with your GCP credentials

# Start the RAG service
npm run dev

API Endpoints

Endpoint	Method	Description
/health/live	GET	Liveness probe
/health/ready	GET	Readiness probe with checks
/v1/query	POST	Semantic search
/v1/generate	POST	RAG-powered answer generation
/v1/corpus	GET	Corpus information

License

MIT