MyAIGist

MyAIGist MCP Server is a local, MCP-compatible knowledge management solution. It processes documents, answers questions with RAG, and maintains a persistent vector store so you can build a multi-document knowledge base entirely on your machine.

How to use

You connect to the MCP Server from any MCP client (Claude Desktop, Cursor, or other compatible clients) and start by processing documents into your knowledge base. Use the document processing tools to add PDFs, DOCX, TXT, and URLs, then ask questions to retrieve answers based on the aggregated content. You can build a unified summary across multiple documents and perform continuous knowledge queries without re-processing the same files.

Practical usage patterns include: processing a single document, sending raw text to add to the knowledge base, crawling a URL to extract content, or batching several files for a combined summary. After adding content, you can pose questions like “What are the main findings?” or “What are the payment terms mentioned across documents?” and receive Claude-powered answers enhanced by OpenAI embeddings for precise retrieval.

How to install

Prerequisites you need before installation include Python 3.8 or newer, an Anthropic API key for Claude text generation, and an OpenAI API key for embeddings. Also ensure you have an MCP-compatible client installed (Claude Desktop, Cursor, or another MCP client).

Step-by-step installation and setup you should follow on your machine:

1. Install dependencies by navigating to the MCP server directory and installing the required Python packages.

cd /path/to/myaigist_mcp
pip install -r requirements.txt

Configure your environment and client

Create a local environment file and add your API keys for Claude and embeddings. This keeps credentials separate from your code while the server runs on your machine.

cp .env.example .env

Edit the new .env file to include:

• ANTHROPIC_API_KEY for Claude
• OPENAI_API_KEY for embeddings

## Connect your MCP client to the server

You configure Claude Desktop or Cursor to connect to the local MCP server so documents you process flow into the knowledge base. The server runs locally and will be ready for use after you start your client.

Example configuration shown for Claude Desktop and Cursor is used to run the server as a local process.

## Start and run details

The server runs as a local process. Start it by launching the Python script from your MCP directory. When your MCP client opens, the server initializes and becomes ready to accept tool commands.

## Code blocks you will configure in clients

The following configuration enables a local Python-based MCP server connection for Claude Desktop and Cursor. Copy this JSON into the appropriate MCP client settings to launch the local server.

{ "mcpServers": { "myaigist": { "command": "python3", "args": ["/path/to/myaigist_mcp/server.py"] } } }

## What happens after you connect

Once connected, you can start processing documents, performing Q&A with RAG, and using the vector store to retrieve relevant content. The vector store persists across sessions, so your knowledge base remains available after restarts.

## Available tools

### process\_document

Process a document from LOCAL FILE PATH (PDF, DOCX, TXT) and add to knowledge base. Parameters: file\_path (string, required), title (string, optional), summary\_level (quick/standard/detailed).

### process\_uploaded\_document

Process a document attached to Claude Desktop. Parameters: content (string, required), filename (string, required), summary\_level (quick/standard/detailed).

### process\_text

Process raw text and add to knowledge base. Parameters: text (string).

### process\_url

Crawl a web URL, extract content, and add to knowledge base. Parameters: url (string).

### process\_batch

Process multiple files and generate a unified summary. Parameters: paths (list of strings).

### ask\_question

Ask questions about stored documents using RAG. Parameters: question (string).

### list\_documents

List all documents in the knowledge base with metadata.

### delete\_document

Delete a specific document by ID.

### clear\_all\_documents

Clear the entire knowledge base.

### get\_status

Get system status and knowledge base statistics.