Kodezi CLI

AI-Powered Terminal Assistant for Intelligent Code Development

Kodezi CLI is an advanced AI-powered terminal assistant that provides intelligent code editing, analysis, and automation directly in your terminal. Transform your development workflow with natural language interactions and powerful built-in tools.

Installation

Install Kodezi CLI globally using npm:

``bash
npm install -g kodezi-cli
`

Verify the installation:

`bash
kodezi --version
`

Features

$3

!Conversation Interface

- Natural Language Interface: Communicate with AI using plain English for all development tasks
- Session Persistence: Maintain conversation context across sessions with SQLite-backed storage
- Real-time Interaction: Streaming responses with immediate feedback and multi-turn conversations
- Automatic Summarization: Compress long sessions while preserving important context

$3

- 12 Built-in Tools: Complete set of utilities for file operations, directory browsing, code search, and command execution
- File Operations: Read, write, edit, and multi-edit files with intelligent content handling
- Directory Management: List, glob pattern matching, and grep search capabilities
- Command Execution: Full shell environment access with real-time output streaming
- Network Operations: HTTP requests and file downloads with progress tracking

$3

- LSP Integration: Real-time code intelligence and diagnostics across multiple programming languages
- Multi-Language Support: Works with any language that has an LSP server
- Error Detection: Automatic identification of warnings and errors in your codebase
- Code Context: Enhanced understanding of project structure and relationships

$3

- MCP Support: Extend functionality using Model Context Protocol servers
- Custom Tools: Add project-specific tools without modifying the CLI
- Multiple Servers: Connect to multiple MCP servers simultaneously
- Auto-Discovery: Automatic tool registration from connected MCP servers

$3

!Permission System

- Permission System: Approval prompts for all file modifications and command executions
- Fine-grained Control: Per-tool, per-directory, and per-action permissions
- Allowlist Support: Pre-approve trusted operations and tools
- Session-level Persistence: Remember permissions for current session

$3

- Non-interactive Mode: Perfect for CI/CD pipelines and scripting
- Auto-approval Options: Skip prompts in trusted environments
- Pipe Support: Accept input from stdin for batch operations
- Exit Code Handling: Proper error reporting for automated workflows

Quick Start

!Interactive Mode

1. Authenticate your account:
`bash
kodezi auth login
`

2. Start interactive mode:
`bash
kodezi
`

3. Begin with simple commands:
`bash
> "What files are in this project?"
> "Show me the contents of package.json"
> "Find all TODO comments in the codebase"
`

4. Exit when done:
`bash
> "exit"
`

Authentication

!Login Process

Kodezi CLI requires authentication to access AI features:

$3


`bash
kodezi auth login
`

You'll be prompted to provide:
- Email: Your Kodezi account email
- API Token: Available from your Kodezi dashboard

$3

`bash
kodezi auth status
`

$3

`bash
kodezi auth logout
`

Commands

Kodezi CLI provides five main commands:

$3

Start an interactive AI session with persistent context:
`bash
kodezi [options]

Options:
-c, --cwd Working directory (default: current directory)
-d, --debug Enable debug mode with detailed logging
-y, --yolo Auto-approve all permissions (use with caution)
`

$3


Execute single tasks without starting a session:
`bash
kodezi run [options] [prompt]

Options:
-q, --quiet Suppress spinner and progress indicators
-y, --yolo Auto-approve all permissions

Examples:

kodezi run "Generate a README for this project"
echo "Explain this code" | kodezi run
`

$3

Manage authentication credentials:
`bash
kodezi auth

Commands:
login Login with email and API token
logout Remove stored credentials
status Show current authentication status
`

$3


Manage session directories and history:
`bash
kodezi dirs [options]

Options:
--list List all session directories
--clean Clean up old session data
`

$3


View and manage application logs:
`bash
kodezi logs [options]

Options:
--tail Follow logs in real-time
--clear Clear existing logs
`

Configuration

Kodezi CLI uses JSON-based configuration stored in your system's config directory. Create or edit the configuration file to customize behavior:

$3


- Windows: %APPDATA%/kodezi-cli/config.json
- macOS: ~/Library/Application Support/kodezi-cli/config.json
- Linux: ~/.config/kodezi-cli/config.json

$3

`json
{
"api": {
"base_url": "https://api.kodezi.com",
"timeout": 30000
},
"permissions": {
"auto_approve": false,
"allowed_tools": ["view", "ls", "grep"],
"blocked_paths": [".env", "secrets/"]
},
"options": {
"context_paths": [
"README.md",
"ARCHITECTURE.md"
],
"max_tokens": 8192,
"temperature": 0.7
}
}
`

$3

- Auto-approve: Skip permission prompts for trusted environments
- Allowed Tools: Pre-approve specific tools to reduce prompts
- Context Paths: Additional files to include for project understanding
- Max Tokens: Control response length and cost
- Blocked Paths: Prevent access to sensitive files

Advanced Features

$3

Enable real-time code intelligence by configuring Language Server Protocol support:

`json
{
"lsp": {
"typescript": {
"command": "typescript-language-server",
"args": ["--stdio"]
},
"python": {
"command": "pylsp",
"args": []
},
"go": {
"command": "gopls",
"args": []
}
}
}
`

Benefits:
- Real-time error and warning detection
- Enhanced code context understanding
- Multi-language diagnostic support
- Background processing without blocking interaction

$3

Extend Kodezi CLI with Model Context Protocol servers:

`json
{
"mcp": {
"database": {
"type": "stdio",
"command": "npx",
"args": ["@kodezi/mcp-database"]
},
"custom-tools": {
"type": "stdio",
"command": "node",
"args": ["/path/to/your-mcp-server.js"]
}
}
}
`

Capabilities:
- Add custom tools for your specific workflow
- Connect to multiple MCP servers simultaneously
- Share tools across team through standardized protocol
- Extend functionality without modifying the CLI

$3

Improve AI understanding by providing additional project context:

`json
{
"options": {
"context_paths": [
"ARCHITECTURE.md",
"CONTRIBUTING.md",
"docs/api.md",
"package.json"
]
}
}
`

The AI automatically accesses these files to better understand your project structure, conventions, and requirements.

Usage Examples

$3


`bash
> "Review the security of auth.js and suggest improvements"
> "Find potential memory leaks in the codebase"
> "Check for unused imports across all TypeScript files"
`

$3

`bash
> "Generate JSDoc comments for all functions in utils.ts"
> "Create API documentation for the routes in server.js"
> "Add inline comments explaining the algorithm in sort.py"
`

$3

`bash
> "Refactor callback functions to async/await in database.js"
> "Convert class components to hooks in React files"
> "Update deprecated APIs in the entire codebase"
`

$3

`bash
> "Generate unit tests for the UserService class"
> "Debug why the authentication tests are failing"
> "Add error handling to the API endpoints"
`

$3

`bash
> "Set up ESLint configuration for this TypeScript project"
> "Add GitHub Actions workflow for CI/CD"
> "Update dependencies and fix breaking changes"
`

$3

`bash

CI/CD Integration

kodezi run --yolo "Run tests and generate coverage report"

Batch Processing

for file in src/*.js; do
kodezi run "Add error handling to $file"
done

Daily Maintenance

kodezi run --quiet "Check for code quality issues and security vulnerabilities"
`

License

This project is licensed under the terms specified in LICENSE.md.

---

Ready to transform your development workflow? Start with npm install -g kodezi-cli` and experience AI-powered coding assistance directly in your terminal.

For detailed documentation, visit app.kodezi.com/docs.