About Intuz

This library is maintained by Intuz, an AI-first software development company specializing in Agentic AI Development and Custom AI Development.

INTUZ is presenting an enterprise knowledge retrieval and RAG MCP server.

Enterprise Knowledge MCP Server

An MCP server that exposes semantic search and RAG over enterprise documents as standardized tools for any MCP-compatible client.

Related MCP server: LTM MCP Server

Use Cases

1. AI-Powered Knowledge Access for Non-Technical Users

• Problem: Business stakeholders cannot query or search internal document repositories without developer help.

• Solution: An MCP server that makes enterprise knowledge available as clean tools — any MCP client (Claude Desktop, etc.) can retrieve and cite information from PDFs, Word documents, and spreadsheets.

• Outcome: Instant, citation-backed answers from company knowledge without touching source systems directly.

2. Automated Document Retrieval and RAG

• Problem: LLMs lack access to internal, proprietary knowledge and hallucinate when asked about it.

• Solution: The server retrieves the most relevant document chunks and assembles a citation-tagged context block ready to drop into any LLM prompt.

• Outcome: Accurate, grounded responses with precise citations pointing to pages, headings, or spreadsheet sheets.

3. Scalable Multi-Source Knowledge Indexing

• Problem: Enterprise knowledge is scattered across SharePoint, Confluence, Notion, Google Drive, Slack, and local file systems.

• Solution: A modular connector architecture — each new data source is one class implementing two methods; nothing downstream changes.

• Outcome: A single unified knowledge index across all enterprise sources, searchable via natural language.

Description

The Enterprise Knowledge MCP Server bridges the gap between enterprise document repositories and AI assistants. It uses FastMCP to expose ingestion, semantic search, and RAG context assembly as standardized tools, so any MCP-compatible client gets read-only, citation-backed access without touching source systems.

This application enables users to:

• Index PDF, DOCX, and XLSX files locally — no document text leaves the machine.

• Perform semantic search with metadata filters (source type, department, author).

• Retrieve citation-tagged RAG context blocks ready for any LLM prompt.

• Extend to new data sources (SharePoint, Confluence, Notion, Google Drive, Slack) by writing one connector class.

Built with FastMCP, ChromaDB, and sentence-transformers, the system is fully local, modular, and production-leaning.

Features

Architecture

[MCP Client — Claude Desktop, etc.]
        |
    (stdio / MCP protocol)
        |
[server.py — FastMCP tools]
        |
[rag/retriever.py — KnowledgeService]
        |
   ┌────┴────┐
   |         |
[ingestion]  [storage/vector_store.py — ChromaDB]
   |
   ├── extractor.py   (PDF pages / DOCX headings+tables / XLSX sheets)
   ├── chunker.py     (sentence-packed overlapping chunks)
   └── embedder.py    (sentence-transformers, local)
        |
[connectors/]
   └── local_files.py  (PDF / DOCX / XLSX — implemented)
   └── ...             (SharePoint / Confluence / Notion / Drive / Slack — roadmap)

Project Structure

enterprise-knowledge-mcp/
├── data/
│   ├── raw/                         # drop source files here
│   └── chroma/                      # Chroma persistent index (auto-created)
├── src/
│   ├── server.py                    # MCP server entry point (FastMCP tools)
│   ├── config.py                    # centralised env-var configuration
│   ├── connectors/
│   │   ├── base.py                  # BaseConnector ABC + SourceDocument dataclass
│   │   └── local_files.py           # local PDF/DOCX/XLSX connector (implemented)
│   ├── ingestion/
│   │   ├── extractor.py             # text extraction per mime type
│   │   ├── chunker.py               # sentence-boundary chunker with overlap
│   │   └── embedder.py              # sentence-transformers wrapper
│   ├── rag/
│   │   └── retriever.py             # KnowledgeService: ingest + search + RAG context
│   └── storage/
│       └── vector_store.py          # Chroma vector store wrapper
├── Screenshots/
│   └── logo.jpg
├── .env.example                     # copy to .env and configure
├── .gitignore
├── requirements.txt
└── README.md

Getting Started

Prerequisites

• Python 3.10+

• pip

Quick Setup (Recommended)

git clone <repository-url>
cd enterprise-knowledge-mcp

# Install dependencies
pip install -r requirements.txt

# Configure environment
cp .env.example .env
# Edit .env if you need non-default paths or a different embedding model

Manual Setup

# Create and activate a virtual environment
python3 -m venv venv
source venv/bin/activate          # Windows: venv\Scripts\activate

# Install dependencies
pip install -r requirements.txt

# Configure environment
cp .env.example .env
# Edit .env as needed (all values have sensible defaults)

Running the Server

python src/server.py

The server runs over stdio — the standard transport for local MCP clients.

Claude Desktop Configuration

Add this block to claude_desktop_config.json:

• macOS: ~/Library/Application Support/Claude/claude_desktop_config.json

• Windows: %APPDATA%\Claude\claude_desktop_config.json

{
  "mcpServers": {
    "enterprise-knowledge": {
      "command": "python",
      "args": ["/absolute/path/to/enterprise-knowledge-mcp/src/server.py"]
    }
  }
}

Replace /absolute/path/to/ with the real path on your machine.

Configuration

Usage

1. Drop PDF, DOCX, or XLSX files into data/raw/.

2. Connect an MCP client (e.g. Claude Desktop) to the server.

3. Use the ingest_local_directory tool to index documents.

4. Ask natural language questions — the client will call search_knowledge or get_rag_context automatically.

Programmatic ingest example:

import sys; sys.path.insert(0, "src")

from connectors.local_files import LocalFileConnector
from rag.retriever import KnowledgeService

svc = KnowledgeService()
result = svc.ingest_connector(LocalFileConnector(root_dir="data/raw"))
print(result)
# {"documents_indexed": N, "chunks_indexed": M, "errors": [...]}

MCP Tools

Dependencies

Testing

# From the repo root with venv activated
pytest -v

Connector Roadmap

Copyright (c) 2026 Intuz Solutions Pvt Ltd. Permission is hereby granted, free of charge, to any person obtaining a copy of this software and associated documentation files (the "Software"), to deal in the Software without restriction, including without limitation the rights to use, copy, modify, merge, publish, distribute, sublicense, and/or sell copies of the Software, and to permit persons to whom the Software is furnished to do so, subject to the following conditions: THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY, FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM, OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE SOFTWARE.