CodeGenie Agent

Still in development, will be deployed soon

An intelligent AI-powered software engineering agent that can analyze, understand, and execute various programming tasks autonomously.

Features

• AI-Powered: Uses Vercel's AI SDK and multi-model support for intelligent task understanding and execution
• Multi-Tool Support: Equipped with various development tools for comprehensive code manipulation
• Secure Operations: Implements robust security measures including path validation and glob pattern sanitization
• TypeScript: Built with modern TypeScript for type safety and better developer experience

Available Tools

Core Development Tools

• Bash Tool (tools/bash.ts): Execute shell commands with timeout protection
• Read Tool (tools/read.ts): Read file contents safely
• Write Tool (tools/write.ts): Create and modify files
• Edit Tool (tools/edit.ts): Perform targeted file edits
• Glob Tool (tools/glob.ts): Search files using glob patterns with security validation
• Grep Tool (tools/grep.ts): Search within file contents

Security Features

• Path Validation (utils/safePath.ts): Prevents directory traversal attacks
• Glob Pattern Sanitization (utils/valGlobPattern.ts): Validates and sanitizes file patterns
• Input Validation: Comprehensive validation using Zod schemas

=� Getting Started

Prerequisites

• Node.js (v16 or higher)
• npm or yarn
• OpenAI API key

Installation

1. Clone the repository:

git clone <repository-url>
cd codegenie/agent

2. Install dependencies:

npm install

3. Set up environment variables: Create a .env file in the agent directory:

OPENAI_API_KEY=your_openai_api_key_here

Usage

Run the agent:

npm start

The agent will prompt you for a task description. Example tasks:

• "Create a new React component for user authentication"
• "Find all TypeScript files with TODO comments"
• "Refactor the utils directory to use ES6 modules"
• "Add unit tests for the validation functions"

Configuration

Environment Variables

• OPENAI_API_KEY: Your OpenAI API key (required) OR - Choose your own model from ai sdk

Tool Limits

• Maximum bash command timeout: 120 seconds
• Maximum glob pattern length: 200 characters
• Maximum glob results: 100 files
• Maximum agent steps: 10 per task

Security

The agent implements multiple security layers:

1. Path Validation: Prevents access to system directories and parent directory traversal
2. Pattern Sanitization: Validates glob patterns to prevent injection attacks
3. Input Validation: Uses Zod schemas to validate all inputs
4. Ignored Directories: Automatically excludes node_modules and dist directories

Dependencies

Core Dependencies

• ai: AI SDK for model integration
• @ai-sdk/openai: OpenAI provider
• zod: Schema validation
• fast-glob: Efficient file pattern matching
• dotenv: Environment variable management

Development Dependencies

• tsx: TypeScript execution
• @types/node: Node.js type definitions
• typescript: TypeScript compiler

Contributing

1. Fork the repository
2. Create a feature branch
3. Make your changes
4. Add tests if applicable
5. Submit a pull request

License

ISC License

=

Examples

Example 1: File Search

User: "Find all TypeScript files in the src directory"
Agent: Uses glob tool to search for `**/*.ts` patterns

Example 2: Code Analysis

User: "Analyze the authentication logic in the codebase"
Agent: Uses grep and read tools to examine auth-related code

Example 3: Refactoring

User: "Convert all require() statements to ES6 imports"
Agent: Uses glob to find JS files, read to examine content, and edit to make changes

Troubleshooting

Common Issues

1. API Key Error: Ensure your OpenAI API key is correctly set in .env
2. Permission Denied: Check file permissions for target directories
3. Pattern Validation Failed: Ensure glob patterns are valid and don't contain dangerous sequences

Debug Mode

The agent logs all tool executions and responses to help with debugging.

---

Built by Shilendra Singh - - - > "Hello guys"