Cursor Chat History Vectorizer & Dockerized Search MCP Server

Overview: What is Cursor Chat History Vectorizer & Dockerized Search MCP Server?

The Cursor Chat History Vectorizer & Dockerized Search MCP Server is a specialized tool designed to make your Cursor chat history searchable for advanced AI workflows. It facilitates the extraction of user prompts from local state.vscdb files within Cursor's workspace storage, converts these prompts into text embeddings via a locally running Ollama instance, and stores them in a LanceDB vector database. The centerpiece is an API server (referred to as an "MCP server") that exposes a simple search endpoint for retrieving similar chat history based on input queries.

🔧 Core Features & MCP Capabilities

This project offers several core features essential for integrating with AI applications via Model Context Protocol (MCP):

1. Data Extraction: Scans specific Cursor workspace storage paths to identify and extract user prompts from state.vscdb files.
2. Embedding Generation: Uses Ollama embeddings (nomic-embed-text:latest) to convert text data into vector representations.
3. Vector Database Storage: Efficiently stores the extracted data, including original text, source file locations, roles (e.g., user vs. assistant), and vector embeddings in a LanceDB database.
4. Dockerized API Implementation: Includes a FastAPI application container that acts as an MCP server for search operations.

AI Application Integration

The MCP server supports compatibility with popular AI applications such as:

• Claude Desktop
• Continue
• Cursor

While full support is available for Claude Desktop, Continue, and Cursor, there are limitations noted for tools within these applications. Specifically, Cursor does not currently have prompt storage or retrieval functionality.

⚙️ MCP Architecture & Protocol Implementation

The implementation leverages Model Context Protocol (MCP) to enable seamless integration between the Cursor Chat History Vectorizer and various AI applications. The core components include:

• Data Extraction Module: Extracts user prompts from Cursor's state.vscdb files.
• Embedding Generation Module: Processes extracted text using Ollama embeddings and stores results in a structured format.
• LanceDB Storage: Holds vectorized data, optimizing for fast query performance through efficient indexing mechanisms.
• MCP API Server (FastAPI): Exposes an MCP-compliant interface allowing external clients to perform search queries against the stored chat history.

MCP Protocol Flow Diagram

The following Mermaid diagram illustrates how the MCP client interacts with the server:

graph TD
    A[AI Application] -->|MCP Client| B[MCP Protocol]
    B --> C[MCP Server]
    C --> D[Data Source/Tool]
    style A fill:#e1f5fe
    style C fill:#f3e5f5
    style D fill:#e8f5e8

This diagram visually represents the communication flow where the AI application client sends requests through MCP, communicates with the server over a protocol (MCP), and retrieves relevant data from connected tools or storage components.

🚀 Getting Started with Installation

System Requirements

• Python 3.6 or higher
• Docker installed on your system
• Ollama instance running locally

Step-by-Step Guide

1. Clone the Repository: Fetch the source code from GitHub.
2. Install Dependencies:

   pip install -r requirements.txt
   

3. Run the Vectorizer Script:

   python vectorizer.py
   

4. Build and Run MCP Server Docker Container:
   • Adjust environment variables in docker-compose.yml as needed.
   • Execute deployment with:

     docker-compose up --build
     

💡 Key Use Cases in AI Workflows

Work Scenario 1: Prompt Generation

This setup enhances AI workflows by enabling more contextual and relevant prompts. For instance, an AI bot can query its own past interactions to generate customized responses based on historical patterns.

Implementation: During the interaction, the bot queries the MCP server with a partial or context-based prompt. The response is filtered from stored conversations, ensuring consistency and relevance in the generated outputs.

Work Scenario 2: Knowledge Base Construction

The vectorized chat history can serve as a knowledge base for AI assistants, where they can retrieve information based on user inputs similar to those found in past interactions.

Implementation: When a new query comes in, the MCP server searches through stored responses to find relevant documents. This ensures that the AI response is directly aligned with historical user concerns and queries.

🔌 Integration with MCP Clients

The Cursor Chat History Vectorizer & Dockerized Search MCP Server integrates seamlessly into several popular AI platforms via MCP support:

• Claude Desktop
• Continue
• Cursor

Integration involves ensuring compatibility with the protocol specifications. Customization may be required depending on specific application needs.

📊 Performance & Compatibility Matrix

The following table outlines the MCP client status for integration in this server setup.

MCP Client	Resources	Tools
Claude Desktop	✅	✅
Continue	✅	✅
Cursor	❌	✅

🛠️ Advanced Configuration & Security

Configuration

Customize the configuration via a .env file or Docker environment variables:

graph TD;
    A[Environment Configuration] --> B[MCP Server Initialization]
    style A fill:#d7e2f0

Example:

API_KEY=your_api_key_here
VECTOR_DATABASE_PATH=/path/to/your/lancedb/database

Security

Ensure secure operation by:

1. Using environment variables: Storing sensitive information like API keys.
2. Securing network connections: Employing TLS encryption for all communication.
3. Role-based access control (RBAC): Implement user roles and permissions to manage data access.

❓ Frequently Asked Questions (FAQ)

Q1: How can I troubleshoot connectivity issues with Ollama?

A1: Verify that your Ollama instance is running locally and accessible from where you run the script or container. Check network configurations, firewall settings, and ensure there are no port conflicts.

Q2: Can this server be used with non-MCP compatible applications?

A2: No direct support for non-compliant applications without significant modifications to the client or server. However, custom scripts can bridge certain gaps temporarily.

Q3: What data protection measures are in place?

A3: The setup supports secure storage and transmission of data through encryption and limited access controls. Ensure all communication is encrypted using TLS/SSL.

Q4: How do I handle large volumes of chat history?

A4: Optimize database performance with indexing, partitioning techniques, and regular maintenance tasks to manage growing datasets efficiently.

Q5: Can I extend this solution for other tools beyond existing clients?

A5: Yes, but custom modifications may be required. The MCP protocol supports extensibility through client-specific adapters or plugins.

👨‍💻 Development & Contribution Guidelines

Contributions are welcome! Developers interested in contributing should:

1. Fork the repository.
2. Create a new branch: git checkout -b feature-branch.
3. Commit your changes: git commit -am 'Your description'.
4. Push to the branch: git push origin feature-branch.
5. Open a pull request.

🌐 MCP Ecosystem & Resources

For more information on Model Context Protocol and related resources, visit:

• Model Context Protocol Official Documentation
• Community forums and mailing lists for ongoing support and discussion.

By participating in this project, you'll be part of a growing ecosystem dedicated to enhancing AI application compatibility through standardized protocols like MCP.