Gemini Api - C# Gemini API Wrapper

Language / 言語

🇯🇵 日本語 | 🇬🇧 English | 🌐 Google Translate

C# reverse-engineered wrapper for Google Gemini web app.

This project aims to create a C# version of the Python project below: https://github.com/HanaokaYuzu/Gemini-API

Base Version:

• Repository: https://github.com/HanaokaYuzu/Gemini-API
• Commit: 3f2b935420fd688e1ab84d937de2c33a871697c0
• Date: 2026-02-10
• Message: Fix the retry logic for unsaved conversations. (#233)

Overview

A C# port of the original Python Gemini API project. Provides an async-first wrapper for the Google Gemini Web API with the following features:

• Persistent Cookies - Automatic cookie management and updates
• Image Generation - Support for image generation and editing
• System Prompts - Customizable prompts using Gems
• Extensions Support - Integration with Gemini extensions
• Streaming Mode - Real-time content generation
• Multi-turn Conversations - Built-in chat session management
• Async - Full .NET async/await support

Requirements

• .NET 8.0 or higher
• C# 11 or higher
• Valid Google account with Gemini access

Dependencies

This library depends on the following NuGet packages:

Package	Version	Purpose
System.Net.Http	4.3.4	HTTP communication
System.Text.Json	8.0.0	JSON serialization/deserialization
HttpClientFactory	1.0.0	HTTP client factory pattern

Note: When using the NuGet package, these dependencies are automatically resolved. If you manually reference the DLL files, you'll need to install these dependencies separately in your project.

Installation

For detailed installation instructions, see INSTALLATION.md.

Quick summary:

• Build from source: git clone and dotnet build
• Download from Release (Recommended): Use local NuGet feed or GitHub Packages

Authentication

1. Access https://gemini.google.com and log in
2. Open Developer Tools (F12) and navigate to the Network tab
3. Refresh the page
4. Click on a request and copy the following cookie values:
   • __Secure-1PSID
   • __Secure-1PSIDTS (optional)

Quick Start

Basic Usage

using GeminiWebApi;
using GeminiWebApi.Utils;

Logger.SetLogLevel(Logger.LogLevel.Info);

var client = new GeminiClient(
    securePsid: "YOUR_SECURE_1PSID",
    securePsidts: "YOUR_SECURE_1PSIDTS"
);

await client.InitializeAsync(timeout: 30);

var response = await client.GenerateContentAsync("Hello World!");
Console.WriteLine(response.Text);

await client.CloseAsync();

Streaming

await foreach (var chunk in client.GenerateContentStreamAsync("Tell me a story"))
{
    Console.Write(chunk.TextDelta);
}

Multi-turn Conversation

var chat = client.StartChat();

var response1 = await chat.SendMessageAsync("What is machine learning?");
Console.WriteLine(response1.Text);

var response2 = await chat.SendMessageAsync("Can you give me an example?");
Console.WriteLine(response2.Text);

Model Selection

using GeminiWebApi.Constants;

var response = await client.GenerateContentAsync(
    "What model are you?",
    model: ModelType.Gemini30Flash
);

Available Models:

• ModelType.Unspecified - Default model
• ModelType.Gemini30Pro - Gemini 3.0 Pro
• ModelType.Gemini30Flash - Gemini 3.0 Flash
• ModelType.Gemini30FlashThinking - Gemini 3.0 Flash Thinking

Gems Panel (System Prompts)

// Fetch available Gems
var gems = await client.FetchGemsAsync(includeHidden: false);

// Search for a specific Gem
var codingGem = gems.Get(name: "Coding Helper");

// Create a custom Gem
var customGem = await client.CreateGemAsync(
    name: "Python Expert",
    prompt: "You are an expert Python developer",
    description: "Helps with Python programming"
);

// Use Gem in conversation
var chat = client.StartChat(gem: customGem.Id);
var response = await chat.SendMessageAsync("How do I use decorators?");

Image Processing

// Generate an image
var response = await client.GenerateContentAsync("Generate an image of a sunset");

// Access images in the response
foreach (var image in response.Images)
{
    Console.WriteLine($"Image: {image.Title}");
    await image.SaveAsync(path: "./images/", filename: "sunset.png");
}

Error Handling

using GeminiWebApi.Exceptions;

try
{
    var response = await client.GenerateContentAsync("Hello");
}
catch (AuthenticationException ex)
{
    Console.WriteLine($"Authentication failed: {ex.Message}");
}
catch (ModelInvalidException ex)
{
    Console.WriteLine($"Invalid model: {ex.Message}");
}
catch (GeminiException ex)
{
    Console.WriteLine($"Error: {ex.Message}");
}

Logging

Control logging output:

using GeminiWebApi.Utils;

Logger.SetLogLevel(Logger.LogLevel.Debug);      // Verbose logging
Logger.SetLogLevel(Logger.LogLevel.Info);       // Info level
Logger.SetLogLevel(Logger.LogLevel.Warning);    // Warnings only
Logger.SetLogLevel(Logger.LogLevel.Error);      // Errors only

API Reference

GeminiClient

The main class for interacting with the Gemini API.

Methods:

• InitializeAsync() - Initialize the client
• GenerateContentAsync() - Generate content from a prompt
• GenerateContentStreamAsync() - Generate content with streaming
• StartChat() - Create a new chat session
• FetchGemsAsync() - Retrieve available Gems
• CreateGemAsync() - Create a custom Gem
• CloseAsync() - Close and clean up the client

ChatSession

Represents a conversation session.

Methods:

• SendMessageAsync() - Send a message in the chat
• SendMessageStreamAsync() - Send a message with streaming
• ChooseCandidate() - Select a specific response candidate

ModelOutput

The result of content generation.

Properties:

• Text - The complete generated text
• TextDelta - New text since the last chunk (for streaming)
• Thoughts - Model's thinking process (when available)
• Images - List of images in the response

Project Structure

GeminiWebApiDotnet/
├── src/GeminiWebApi/
│   ├── Constants/          # API constants and endpoints
│   ├── Exceptions/         # Custom exceptions
│   ├── Models/             # Data models
│   ├── Utils/              # Utility functions
│   └── GeminiClient.cs     # Main client class
├── example/                 # Sample usage
├── tests/                   # Unit tests
├── GeminiWebApi.csproj     # Project file
└── ReadMe.md

License

AGPL-3.0 License - See the LICENSE file for details

Note

This project is a reverse-engineered implementation and is not affiliated with Google. Use at your own risk and in compliance with Google's Terms of Service.

See Also

For Japanese documentation, see ReadMe.ja.md