Bmorphism Krep MCP Server

🚀 Krep MCP Server

Krep MCP Server is a high-performance string search tool based on the Model Context Protocol (MCP). It offers a unified interface for efficient searching in files or strings.

🚀 Quick Start

📦 Installation

Installing the Krep MCP Server is straightforward:

1. Clone this repository.
2. Navigate to the krep-mcp-server directory.
3. Run the main program using Node.js.

Example Installation Commands

git clone https://github.com/your-username/krep-mcp-server.git
cd krep-mcp-server
node src/index.js

✨ Features

The Krep MCP Server provides the following key features:

1. Unified Interface: Achieve file search, string search, and counting functions through a single function.
2. High-Performance Algorithms:
   • Automatically select the optimal algorithm (KMP, Boyer-Moore-Horspool, Rabin-Karp) based on the pattern length.
   • Support SIMD instruction acceleration.
3. Multithreading Support: Automatically utilize all CPU cores to enhance search efficiency.

💻 Usage Examples

Request Format

<use_mcp_tool>
<server_name>krep</server_name>
<tool_name>krep</tool_name>
<arguments>
{
  "pattern": "search pattern",
  "target": "file path or string to search",
  "mode": "file|string|count",
  "caseSensitive": true|false,
  "threads": null // Automatically detect the number of CPU cores, default is automatic mode
}
</arguments>
</use_mcp_tool>

Parameter Explanation

• pattern (Required): The pattern string to search for.
• target (Required): The file path or the string to be searched.
• mode (Optional):
  • file (Default): Search in a file.
  • string: Search in a string.
  • count: Only count the number of occurrences.
• caseSensitive (Optional): Whether to perform a case-sensitive search, default is true.
• threads (Optional): The number of threads to use, default is to automatically detect the number of CPU cores.

For specific usage examples, please refer to examples.md.

🔧 Technical Details

The Krep MCP Server is designed with a focus on performance:

1. Algorithm Selection:
   • Use the KMP algorithm for extremely short patterns (<3 characters).
   • Apply the Boyer-Moore-Horspool algorithm for medium-length patterns (3 - 16 characters).
   • Employ the Rabin-Karp algorithm for long patterns (>16 characters).
2. Hardware Acceleration: Utilize SIMD instructions to improve performance.
3. Dynamic Multithreading: Automatically utilize all available CPU cores.

Integration

Integration with Cline VSCode Plugin

The Krep MCP Server can be seamlessly integrated with the Cline VSCode plugin, enabling high-performance search functionality directly within the VSCode environment.

Installation Steps

# Install the plugin
./install-cline-integration.sh

# Install before testing the plugin
./run-cline-test.sh

# Run when uninstalling the plugin
./uninstall-cline-integration.sh

Usage in Cline

After integration, you can directly use krep in Cline conversations:

/krep krep pattern="function" target="/path/to/search" mode="file"

For more installation and usage details, please refer to:

• CLINE_SETUP.md - Installation Guide
• CLINE_README.md - Usage Instructions

Integration into the Infinity Topos Environment

The Krep MCP Server is designed with a seamless integration solution for the Infinity Topos environment:

• Babashka Integration: Used to process search results.
• Say Integration: Announce search results audibly through the Say MCP Server.
• Coin-Flip Integration: Determine the search strategy using randomization.

Development Environment

Environment Variables

• CLAUDE_MCP: Set to "true" to enable MCP mode.
• KREP_PATH: Specify the path to the krep executable file.
• DEBUG: Set to "true" to enable detailed log output.
• KREP_TEST_MODE: Set to "true" to enter test mode and use simulated responses.
• KREP_SKIP_CHECK: Set to "true" to skip checking the existence of krep.

HTTP Server Mode

In non-MCP mode, the server starts the following endpoints:

1. /api/search:
   • The POST request body should contain search parameters.

📄 License

This project is licensed under the MIT License.

Contributors

Thank you to all contributors!

Links

• GitHub Repository
• Issue Reporting