ETABS Documentation Assistant MCP Server (Local Embeddings)

This project provides a Model Context Protocol (MCP) server that allows AI models (like Anthropic's Claude via Claude Desktop) to perform semantic searches on user-provided ETABS documentation. It uses local sentence transformer models for embedding generation (via @xenova/transformers.js) and ChromaDB for vector storage, making it free to run after initial setup.

Core Functionality:

• Exposes an MCP tool named search_etabs_docs.
• Accepts a natural language query about ETABS.
• Finds the most relevant sections in the user's indexed ETABS documentation using semantic similarity.
• Returns text snippets from the documentation (with source file references) to the MCP client (e.g., Claude Desktop) to be used as context by the AI model.

🚨 IMPORTANT DISCLAIMERS 🚨

• NO ETABS DOCUMENTATION INCLUDED: This repository DOES NOT contain any ETABS documentation files. ETABS documentation is proprietary software owned by Computers & Structures, Inc. (CSI).
• USER MUST PROVIDE DOCUMENTATION: To use this server, you MUST have a legally obtained copy of the ETABS documentation, typically in .chm format.
• NO AFFILIATION: This project is not affiliated with, endorsed by, or sponsored by Computers & Structures, Inc. (CSI).
• USE AT YOUR OWN RISK: This software is provided "as-is" without warranty. Ensure you comply with all relevant software licenses for ETABS and its documentation.

Architecture Overview

This project consists of two main parts:

1. Python Indexer (index_chm_py/): A script that extracts content from your .chm ETABS documentation file, converts it to text, chunks it, generates embeddings locally using Sentence Transformers, and stores everything in a local ChromaDB database. This needs to be run once initially.
2. Node.js MCP Server (src/): The actual MCP server that runs persistently. It receives search queries via the search_etabs_docs tool, generates an embedding for the query locally, queries the ChromaDB database for similar chunks, and returns the results to the connected MCP client.

graph LR
    U[User] --> Client[MCP Client e.g., Claude Desktop]
    Client -- MCP (stdio) --> Server[Node.js MCP Server]
    Server -- HTTP --> DB[(ChromaDB via Docker)]
    User -- Provides --> CHM[ETABS .chm File]
    CHM -- Used by --> Indexer[Python Indexer Script]
    Indexer --> DB

Prerequisites

Before you begin, ensure you have the following installed:

• Node.js: Version 18 or later recommended (Download).
• npm: Usually included with Node.js.
• Python: Version 3.9 or later recommended (Download). Ensure python and pip are in your PATH.
• Docker: Required to easily run the ChromaDB vector database (Download Docker Desktop). Ensure the Docker daemon/service is running.
• ETABS Documentation File: Your legally obtained etabs.chm (or similarly named) file.
• CHM Extraction Tool: An external command-line tool capable of extracting .chm file contents. This script attempts to use 7z (from 7-Zip) or chmextract.
  • Windows: Install 7-Zip. Ensure 7z.exe is added to your system's PATH environment variable during or after installation.
  • macOS: Install via Homebrew: brew install p7zip chmextract (provides both 7z and chmextract).
  • Linux (Debian/Ubuntu): sudo apt update && sudo apt install p7zip-full libchm-bin (provides 7z and chmextract).

(Verify the tool is callable from your terminal before proceeding).

Setup Instructions

Clone the Repository:

git clone https://github.com/<your-github-username>/etabs-mcp-server-local-embeddings.git
cd etabs-mcp-server-local-embeddings

(Replace <your-github-username> with your actual username)

Install Node.js Dependencies:

`

... View full README on GitHub