Geminimcp

The Gemini MCP Server is a single executable file developed in Go, integrating the Model Control Protocol server of Google Gemini API. It provides dynamic model access, advanced context caching, intelligent file handling, and enhanced search functions, supporting various application scenarios such as code analysis, creative writing, and factual research.

Artificial intelligence chatbots Developer tools #Model Control #Intelligent Caching #File Handling #Search Enhancement .Go

rating : 2 points

downloads : 9

update time : 2025-04-29

What is the Gemini MCP Server?

The Gemini MCP Server is a tool integrating Google Gemini API, aiming to provide powerful language model capabilities through the Model Control Protocol (MCP). It supports automatic update of the latest models, efficient caching, file handling, and advanced context management.

How to use the Gemini MCP Server?

Simply start the server and configure the API key to start using. Through simple commands or client configuration, you can quickly analyze code, perform search tasks, or engage in creative writing.

Applicable Scenarios

Suitable for scenarios such as programming code analysis, complex problem solving, creative writing, and knowledge queries based on search.

Main Features

Dynamic Model AccessThe server automatically fetches the latest Gemini models at startup, ensuring the use of the latest version at all times.

Efficient Context CachingImprove the efficiency of repeated queries through the TTL - controlled caching mechanism.

Intelligent File HandlingSupports multiple file formats with automatic MIME detection and verification functions.

Production - Grade ReliabilityEquipped with error handling, retry mechanisms, and degradation modes to ensure stable operation.

Comprehensive Capability SupportCovers code analysis, general queries, and search tasks with a basis.

Advantages and Limitations

Advantages

A single self - contained binary file without the need for a package manager.

Dynamically update models to always stay up - to - date.

Support efficient caching and file handling.

Have the high reliability and stability required for production environments.

Support multi - scenario tasks such as code analysis, search, and creative writing.

Limitations

Requires a Google Gemini API key.

Some advanced features may be restricted by the API.

For large - scale deployment, performance optimization may be required.

How to Use

Installation and Configuration

After downloading the source code, use Go to compile and generate a single binary file, and set the necessary environment variables.

Client Configuration

Add the server to any MCP - compatible client, such as Claude Desktop.

Execute Tasks

Perform code analysis, search, or creative writing tasks through the configured tools.

Usage Examples

Code Analysis ExampleAnalyze concurrency issues in Go code.

Search Task ExampleFind the latest progress on quantum computing.

Creative Writing ExampleGenerate a story about space exploration.

Frequently Asked Questions

How to obtain a Gemini API key?

How to ensure the cache is effective?

Does it support file upload?

Related Resources

GitHub Repository

Project source code and documentation.

Official Documentation

Detailed introduction to the MCP protocol.

Claude Desktop Configuration Guide

How to configure the Claude Desktop client.

🚀 Libra Gemini MCP Project Documentation

This project provides a comprehensive solution for managing and interacting with Google Gemini models, offering a range of features to enhance efficiency and reliability in various tasks.

🚀 Quick Start

Install Dependencies

go get golang.org/x/oauth2@git://github.com/golang/oauth2/cmd/...

Configure Environment Variables

Property	Details
`GEMINI_API_KEY`	Google Gemini API key. Required
`GEMINI_MODEL`	Default model ID for `gemini_ask`. Default: `gemini-1.5-pro`
`GEMINI_SEARCH_MODEL`	Default model ID for `gemini_search`. Default: `gemini-2.0-flash`
`GEMINI_SYSTEM_PROMPT`	System prompt for general queries. Customized review prompt
`GEMINI_SEARCH_SYSTEM_PROMPT`	System prompt for search. Customized search prompt
`GEMINI_MAX_FILE_SIZE`	Maximum upload file size in bytes. Default: `10485760` (10MB)
`GEMINI_ALLOWED_FILE_TYPES`	Allowed file types, separated by commas. Common text/code types

Run Commands

# Initialize Gemini MCP
go run main.go

# Start the service
./start_service.sh

✨ Features

Core Features

Model Management: Supports multiple Gemini models, including gemini-1.5-pro, gemini-2.0-flash, etc.
Context Caching: Enables context caching with a default storage time of 1 hour.
Thinking Mode: Supports detailed thinking processes, suitable for complex reasoning tasks.
File Handling: Supports uploading common text and code files up to 10MB.

Advanced Features

Search Integration: Built-in Google search function with source annotation support.
Error Recovery: Automatically enters safe mode in case of initialization failure.
Retry Mechanism: Uses an exponential backoff strategy to ensure reliable API communication.
Logging: Provides structured logs for easy troubleshooting.

📚 Model Management

Model Priority

Recommended Models:
- Thinking tasks: gemini-2.5-pro
- Caching tasks: gemini-2.0-flash
- Search tasks: gemini-2.0-flash
Version Selection:
- Prioritize new versions to ensure the use of the latest features and performance optimizations.
- Optimize for specific scenarios and recommend the best model for different tasks.

🛠️ Configuration Optimization

Key Environment Variables

Property	Details
`GEMINI_TIMEOUT`	API timeout in seconds. Default: `90`
`GEMINI_MAX_RETRIES`	Maximum number of retries. Default: `3`
`GEMINI_ENABLE_CACHING`	Whether to enable context caching. Default: `true`
`GEMINI_THINKING_MODE`	Whether to enable the detailed thinking mode. Default: `true`

Performance Tuning

Reduce Timeout: In high-latency environments, shorten the timeout time appropriately.
Adjust Retry Strategy: Adjust the maximum number of retries based on network stability.
Optimize Caching: Set the cache expiration time reasonably to balance hit rate and freshness.

💻 Development Guide

Project Structure

gemini-mcp/
├── cmd/                 # Command-line tools
│   ├── gemini_ask.go    # Conversation interface
│   └── gemini_search.go # Search interface
├── pkg/                 # Core packages
│   ├── model_manager.go  # Model manager
│   ├── caching.go       # Caching module
│   └── search_engine.go  # Search engine
└── README.md             # Project documentation

Development Steps

Initialize the project:

go mod init your_project_name
go mod tidy

Run tests:
```
go test ./...
```
Code inspection:
```
go lint ./...
```
Format code:
```
go fmt ./...
```

⚠️ Runtime Notes

Resource Management

Memory Limit: Monitor cache usage to prevent memory overflow.
Disk Space: Regularly clean up expired caches to avoid running out of storage space.

Security Measures

Key Protection:
- Encrypt the API key in the environment variables.
- Avoid writing the key into the version control system.

🆕 Recent Changes

Improved Model Management: Optimized the multi-model switching logic and improved performance.
Enhanced Caching Mechanism: Introduced a more intelligent caching strategy to reduce redundant calculations.
Refined Thinking Mode: Added more reasoning options to support complex scenarios.
Fixed Known Issues:
- Fixed boundary conditions in file handling.
- Resolved the deadlock issue under the retry mechanism.