Kagi MCP - Intelligent Search Assistant

Kagi MCP is an intelligent assistant based on the Model Control Protocol (MCP) that integrates Kagi's AI services to provide high-quality search-driven conversations, webpage summaries, and text translation capabilities. This tool leverages multiple advanced AI models, automatically selecting the most appropriate model for different task types to ensure optimal response quality.

Features

1. Intelligent Conversation (kagi_chat)

Automatically triggered search-driven intelligent assistant that can:

• Answer almost any question with the latest, most accurate information
• Search for technical documentation and tutorials
• Troubleshoot common errors and issues
• Find configuration guides for software and tools
• Search for code examples of specific functionality
• Get recommended development best practices
• Stay updated on the latest developments in technology and industries

2. Webpage Summarization (kagi_summarize)

Quickly analyze and summarize webpage content:

• Extract main content from long articles
• Extract key information from webpages
• Get main points and conclusions from articles
• Save reading time
• Analyze technical documentation and open source projects
• Extract key findings from papers and research reports

3. Text Translation (kagi_translate)

High-quality text translation tool:

• Translate text from one language to another
• Maintain the original meaning and tone
• Suitable for various types of text, including technical documentation, literary works, etc.
• Support translation between multiple languages

Intelligent Model Selection

The system automatically selects the most appropriate AI model based on the task type:

Conversation Model Selection

• General Knowledge: Optimized for general questions (claude-3-7-sonnet)
• Advanced Reasoning: For complex reasoning problems (o3)
• Balanced Performance: Provides balance between speed and quality (claude-3-7-sonnet)
• Creative Content: For creative content generation (gemini-2-5-pro)
• Technical Analysis: For technical analysis questions (o3)
• Architecture Design: For system architecture discussions (o3)
• Quick Response: For quick factual questions (o4-mini)
• Code Generation: For programming and code-related questions (gemini-2-5-pro)
• Scientific Research: For scientific research questions (gemini-2-5-pro)

Summary Model Selection

• Standard Summary: Provides balanced and detailed content summary (claude-3-7-sonnet)
• Comprehensive Analysis: Delivers in-depth analysis and insights (gemini-2-5-pro)
• Efficient Overview: Quickly provides a concise overview of key points (o4-mini)
• Technical Breakdown: Analyzes technical content in detail (o3)
• Research Summary: Professionally summarizes academic or scientific content (gemini-2-5-pro)

Translation Model Selection

• Standard Translation: Provides high-quality translation for general text (claude-3-7-sonnet)
• High Accuracy: Delivers precise translation for formal documents (gemini-2-5-pro)
• Technical Translation: Accurate translation of professional or technical content (o3)
• Quick Translation: Fast translation for simple content (o4-mini)
• Creative Translation: Preserves style, tone, and creative elements of the original text (gemini-2-5-pro)

Installation and Setup

Prerequisites

• Python 3.8 or higher
• Kagi account and valid Cookie

Installation Steps

1. Clone the repository or download the source code

2. Install dependencies

   pip install -r requirements.txt
   

3. Configure environment variables Create a .env file in the project root directory and add your Kagi Cookie:

   KAGI_COOKIE=your_kagi_cookie_here
   

   How to get your Kagi Cookie:

   • Log in to the Kagi website (https://kagi.com)
   • Open browser developer tools (F12)
   • Go to the Network tab
   • Refresh the page
   • Find any request and look for the Cookie value in Headers
   • Copy the entire Cookie string

Usage

Starting the Service

python kagi.py

Using Commands

• Regular questions will automatically use the kagi_chat tool
• Use the /summarize [URL] command to summarize webpage content
• Use the /translate [text] command to translate text

Examples

1. Regular conversation (automatically uses kagi_chat):

   How to process JSON data in Python?
   

2. Webpage summarization:

   /summarize https://en.wikipedia.org/wiki/Artificial_intelligence
   

3. Text translation:

   /translate Python is a programming language that lets you work quickly and integrate systems more effectively. Chinese
   

Advanced Features

Caching Mechanism

The system implements a caching mechanism that can cache responses in non-session mode, improving response speed and reducing API calls.

Session Management

The system automatically manages session context to maintain conversation coherence. You can start a new conversation by setting new_conversation=True.

Custom Configuration

You can customize API configuration by modifying the KagiConfig class, such as timeout, user agent, etc.

Technical Architecture

This project is built on the following technologies:

• MCP (Model Control Protocol): For building and managing AI tools
• FastMCP: Fast implementation of MCP for creating AI services
• Kagi API: Provides high-quality AI responses and search capabilities
• Requests: For HTTP requests
• HTML2Text: For converting HTML to Markdown format
• Python-dotenv: For environment variable management

Important Notes

• A valid Kagi Cookie is required to use the service
• Cookies have a limited validity period and need to be updated after expiration
• Please comply with Kagi's terms of use and limitations when using the API

Contributions and Feedback

Contributions, issue reports, and feature requests are welcome. Please submit your feedback and contributions through GitHub Issues or Pull Requests.

License

MIT License