Code Assistant

A CLI tool built on Rust to assist with code-related tasks, supporting autonomous exploration of the codebase, file reading and writing, working memory management, file summarization, and interactive communication, and can run as an MCP server.

Developer tools Artificial intelligence chatbots #Code Assistance #MCP Service #LLM Integration #Development Tool .Rust

rating : 2.5 points

downloads : 62

update time : 2025-04-28

What is the Model Context Protocol (MCP) Server?

The MCP server is a tool based on the Model Context Protocol that allows large language models (LLMs) to interact with the local code environment. It can read and write files, manage code context, and perform code-related tasks.

How to use the MCP server?

Through simple command-line configuration and startup, the MCP server can be integrated into the existing development environment, supporting multiple programming languages and frameworks.

Applicable Scenarios

The MCP server is very suitable for developers and teams that need automated code analysis, document generation, or complex task processing.

Main Features

Code Context ManagementEfficiently load and unload file contents for quick retrieval and modification.

File Operation ToolSupports functions such as file reading, writing, and search-and-replace.

Autonomous Exploration AbilityIntelligently explore the codebase to build an understanding of the project structure.

Multi-Model SupportCompatible with multiple large language models, such as Anthropic and OpenAI.

Advantages and Limitations

Advantages

Improve development efficiency

Support multiple programming languages and frameworks

Easy to integrate into the existing workflow

Powerful file operation and code analysis capabilities

Limitations

Requires installation of a specific development environment

Some advanced features may require additional configuration

There is still room for improvement in performance optimization for large codebases

How to Use

Install Code Assistant

Ensure that the Rust toolchain is installed and clone the code repository.

Compile the Project

Run the build command in the project directory.

Start the MCP Server

Use the server command to start the MCP server.

Usage Examples

Example 1: List all API interfacesUse Code Assistant to list all API interfaces in the current project.

Example 2: Optimize Database QueriesOptimize database query statements through Code Assistant.

Frequently Asked Questions

How to install Code Assistant?

Which language models does the MCP server support?

How to record and playback sessions?

Related Resources

Code Assistant GitHub Repository

Get the latest version and documentation.

Model Context Protocol Official Website

Learn detailed information about the MCP protocol.

Claude Desktop

A development tool integrating the MCP server.

🚀 Code Assistant Project Documentation

The Code Assistant is a powerful tool designed to assist developers in analyzing, optimizing, and understanding codebases. It supports multiple AI providers and can operate in two modes: task mode and server mode.

🚀 Quick Start

Task Mode

In task mode, the Code Assistant can execute various analysis tasks. Here are some examples:

# Analyze the codebase in the current directory
code-assistant --task "Explain the purpose of this codebase"

# Analyze a specific directory using OpenAI
code-assistant -p open-ai --path ./my-project -t "List all API endpoints" -v

# Perform code complexity analysis using Google Vertex AI
code-assistant -p vertex --model gemini-1.5-flash -t "Analyze code complexity"

# Find TODO comments using the Ollama model
code-assistant -p ollama -m codellama --task "Find all TODO comments"

# Write documentation using the AI Core provider
code-assistant -p ai-core --task "Document the public API"

# Find performance bottlenecks using the working memory mode
code-assistant --task "Find performance bottlenecks" --agent-mode working_memory

# Continue the previously interrupted task
code-assistant --continue-task

# Launch the graphical user interface
code-assistant --ui

# Record a session for playback
code-assistant --task "Optimize database queries" --record ./recordings/db-optimization.json

# Play back the recorded session at an accelerated speed
code-assistant --playback ./recordings/db-optimization.json --fast-playback

Server Mode

In server mode, the Code Assistant can run as a Model Context Protocol (MCP) server:

code-assistant server [OPTIONS]

✨ Features

AI Provider Support: The supported AI providers include Anthropic, OpenAI, Vertex AI, OpenRouter, and AI Core.
Multi - language Support: Users can choose different models for analysis according to their needs.
File Operations: It supports operations such as file search, replacement, and command execution.
Session Recording: Sessions can be recorded for subsequent playback, and a fast - playback mode is available.

📦 Installation

Environment Variable Configuration

ANTHROPIC_API_KEY: Required when using the Anthropic provider.
OPENAI_API_KEY: Required when using the OpenAI provider.
GOOGLE_API_KEY: Required when using the Vertex provider.
OPENROUTER_API_KEY: Required when using the OpenRouter provider.
PERPLEXITY_API_KEY: Required when using the Perplexity search API tool.

📚 Documentation

Roadmap

Here are some potential future development focuses:

Security: Provide a more secure sandbox environment to restrict access to project files.
UI Improvement: Enhance the user experience of the text input interface by adding Markdown support and code block syntax highlighting.
Agent Improvement: Optimize the working memory mode and message history mode, and explore methods for automatically pruning redundant information.
Fuzzy Matching Search: Research more efficient file search algorithms to improve matching success rates.