Ogoldberg Gemini Context MCP Server

🚀 Gemini Context MCP Server

A powerful implementation of the MCP (Model Context Protocol) server that leverages Gemini's capabilities for context management and caching. This server maximizes the value of Gemini's 2M token context window and provides tools for efficiently caching large contexts.

🚀 Quick Start

Prerequisites

• Node.js (version 14 or higher is recommended)
• npm (version 6 or higher is recommended)
• A Gemini API key

Installation Steps

1. Clone the project repository:

git clone [Repository URL]
cd gemini-context-mcp-server

2. Install dependencies:

npm install

3. Configure environment variables by creating a .env file:

GEMINI_API_KEY=Your_API_Key

4. Start the server:

node index.js

✨ Features

Context Management

• Support for up to 2M token context window - Fully utilize Gemini's context capabilities.
• Text generation - Generate text with context.
• Get context - Retrieve the current context of a specific session.
• Clear context - Clear the context of a particular session.
• Add context - Add specific context entries.
• Search context - Perform semi-content-based context search.

Cache Management

• Create cache - Store large chunks of context information.
• Generate with cache - Generate text using the context in the cache.
• List caches - View all available caches.
• Update cache TTL - Adjust the cache's time-to-live.
• Delete cache - Remove caches that are no longer needed.

📚 Documentation

Core Features

• Context management - Handle context information for multiple sessions, supporting addition, query, and deletion operations.
• Caching mechanism - Provide an efficient context storage solution to optimize the performance of repeated requests.
• Automatic expiration - Cache entries will automatically expire based on the set timestamp to ensure data freshness.

Advanced Features

• Context search - Find relevant context through content similarity to improve information retrieval efficiency.
• Rate limiting - Control the request frequency to prevent service overload.
• Logging - Record operation logs for easy debugging and monitoring.

💻 Usage Examples

Command-line Tools

# Create a new context entry
mcp create-context --session-id 12345 --content "Welcome to Gemini context management"

# Load an existing context
mcp load-context --session-id 12345

# Clear the context of a specific session
mcp clear-context --session-id 12345

Cache Operation Examples

# Create a new cache entry
mcp create-cache --cache-id user-logs --content "/var/log/app"

# Generate text using the cache
mcp generate-with-cache --cache-id user-logs --prompt "Explain the content of the log file"

# Update the cache TTL
mcp update-cache-ttl --cache-id user-logs --ttl 3600

# Delete a cache entry
mcp delete-cache --cache-id user-logs

Configuration Options

Create a .env file with the following options:

GEMINI_API_KEY=Your_API_Key
GEMINI_MODEL=g4 (Optional, default is g4)
API_RATE_LIMIT=100 (Optional, default is unlimited)
CONTEXT_EXPIRATION_MINUTES=60 (Optional, default is 60 minutes)

Further Reading

• For detailed MCP protocol specifications, refer to Documentation.
• To explore more integration examples and advanced usage, visit Official Documentation.

Future Improvements

• Introduce database persistence to ensure that context and cache information is not lost upon restart.
• Implement intelligent caching strategies, such as the LRU replacement algorithm, to improve caching efficiency.
• Integrate a vector database to enhance the semantic search capabilities of the context.

📄 License

This project is licensed under the MIT License.