AgenticCLI V2.0 Beta

AI-powered command line tool with support for Google Gemini 2.5 Flash Lite and AWS Bedrock Nova models.

Beta Notice: This is a beta release. Features are stable but may be subject to changes based on user feedback.

Features

Multi-Model AI Support - Google Gemini 2.5 Flash Lite and AWS Bedrock Nova
Real-time Command Filtering - Type / and see commands filter as you type
Arrow Key Navigation - Navigate through filtered commands with ↑/↓ arrows
Tab Completion - Press Tab to auto-complete selected commands
Smart Input Detection - Automatically detects AI queries vs system commands
Dynamic Status Indicators - Real-time display of system, model, and API status
Application Launcher - Open Excel, Word, Slack, Chrome, VS Code, and more directly
Conversation History - Maintains context across your session
File Operations - Read, analyze, and explain files with AI
Configurable Preferences - Customize temperature, tokens, and behavior
Cross-platform - Works on Windows & Linux
Agent-themed Messages - Friendly "Agent is thinking..." and goodbye messages

Screenshots

Welcome Screen with Status Indicators

AI Queries and System Commands

Help Command - All Available Commands

Development Environment

Welcome Screen

When you start AgenticCLI, you'll see a professional yellow ASCII art banner with real-time status indicators:

Status Indicators

The welcome screen shows three dynamic status indicators:

● Online (Green) - AgenticCLI is running and ready
Model Status:
- ● Ready (Green) - AI model initialized successfully
- ● Error (Red) - Model failed to initialize
- ● Loading (Yellow) - Model is initializing
API Status:
- ● Connected (Green) - API keys are configured
- ● No Keys (Red) - Missing API keys in .env file

Installation

Clone the repository
Install dependencies:
```
npm install
```
Create .env file (copy from .env.example):
```
cp .env.example .env
```

Add your API keys to .env:

# For Google Gemini
GEMINI_API_KEY=your_gemini_api_key_here

# For AWS Bedrock
AWS_ACCESS_KEY_ID=your_aws_access_key
AWS_SECRET_ACCESS_KEY=your_aws_secret_key
AWS_REGION=us-east-1

Link globally:
```
npm link
```
Run from anywhere:
```
agenticcli
```

Command Filtering & Navigation

AgenticCLI features real-time command filtering for easy command discovery:

How to Use

Type / - Shows all available commands below your input
Type /h - Instantly filters to commands containing 'h': /help, /history, /clear-history, /save-chat
Type /r - Filters to: /read, /clear, /clear-history, /reset
Type /c - Filters to: /clear, /config, /clear-history, /save-chat
Use ↑/↓ arrows - Navigate through filtered commands
Press Tab - Auto-complete the selected command
Press Enter - Execute the selected command

Example

agenticcli> /h

> /help
  /history
  /clear-history
  /save-chat

The first matching command is automatically selected (marked with >), making it quick to execute common commands.

Filtering in Action

# Type /r to see read-related commands
agenticcli> /r

> /read
  /reset
  /clear
  /clear-history

# Use arrow keys to select /read, press Tab to complete
agenticcli> /read 

# Type /c to see clear/config commands
agenticcli> /c

> /clear
  /config
  /clear-history
  /save-chat

Commands

CLI Commands

/help - Show all available commands
/exit or /quit - Exit the tool (shows "Agent signing off. Goodbye! 😊")
/clear - Clear screen
/models - List available AI models
/use <model> - Switch AI model (gemini or bedrock)
/status - Show current configuration

History Commands

/history - Show conversation history
/clear-history - Clear conversation context
/save-chat <filename> - Export chat to file

File Commands

/read <file> - Load file into context
/analyze <file> - AI analyzes the file
/explain <file> - AI explains what the file does

Software Commands

excel - Open Microsoft Excel
word - Open Microsoft Word
slack - Open Slack
chrome - Open Chrome Browser
code - Open Visual Studio Code
kiro - Open Kiro IDE
notepad++ - Open Notepad++
git-bash - Open Git Bash

Configuration Commands

/config - Show current settings
/set <key> <value> - Update setting
/reset - Reset to defaults

System Commands

Just type regular commands like: ls, pwd, cd, cat, etc.

Usage Examples

Command Filtering

# Show all commands
agenticcli> /
> /help
  /exit
  /quit
  /clear
  /models
  ... (all commands shown)

# Filter commands as you type
agenticcli> /h
> /help
  /history
  /clear-history
  /save-chat

# Use arrows to select, Tab to complete, Enter to execute
agenticcli> /help

Executing System Commands

agenticcli> ls
Documents  Downloads  Pictures  Projects

agenticcli> pwd
/home/user/projects

agenticcli> cat package.json
{
  "name": "my-project",
  ...
}

AI Queries

When you ask a question, the agent shows a thinking message and responds:

agenticcli> what is nodejs?

Agent is thinking...

Node.js is a JavaScript runtime built on Chrome's V8 JavaScript engine. 
It allows you to run JavaScript on the server side, enabling you to build 
scalable network applications. Key features include:

- Event-driven, non-blocking I/O model
- NPM package ecosystem
- Cross-platform compatibility
- Built-in modules for file system, HTTP, etc.

agenticcli> explain async/await in javascript

Agent is thinking...

Async/await is a modern way to handle asynchronous operations in JavaScript...

The green "Agent is thinking..." message appears while the AI processes your query.

Switching AI Models

agenticcli> /models

Available Models:

  gemini [ACTIVE]
  bedrock

agenticcli> /use bedrock
Success: Switched to bedrock

agenticcli> /status

Current Status:

Model: bedrock
Temperature: 0.7
Max Tokens: 2048
Auto-execute safe commands: false
History entries: 12

File Operations

# Read a file into context
agenticcli> /read app.js
Success: Loaded app.js (2048 bytes) into context

# Ask AI about the file
agenticcli> what does this file do?

Agent is thinking...

This file is the main entry point for your application. It sets up an 
Express server, configures middleware, and defines routes...

# Analyze a file
agenticcli> /analyze src/utils/helper.js

Agent is thinking...

Code Analysis for helper.js:

Strengths:
- Well-structured utility functions
- Good error handling
- Clear function names

Suggestions:
- Add JSDoc comments for better documentation
- Consider adding input validation
- Extract magic numbers into constants

# Explain a file
agenticcli> /explain config.json

Agent is thinking...

This configuration file defines settings for your application:
- port: 3000 - The server will listen on port 3000
- database: Connection settings for MongoDB
- logging: Enables debug mode logging
...

Saving Conversations

agenticcli> /save-chat my-session.txt
Success: Chat saved to: /home/user/.agenticcli-history/my-session.txt

agenticcli> /history

Conversation History:

[User]: what is nodejs?
[Assistant]: Node.js is a JavaScript runtime...

[User]: /read app.js
[Assistant]: [File loaded]

[User]: what does this file do?
[Assistant]: This file is the main entry point...

Opening Applications

Launch your favorite applications directly from AgenticCLI:

# Open Microsoft Office applications
agenticcli> excel
agenticcli> word

# Open development tools
agenticcli> code .
agenticcli> git-bash
agenticcli> kiro

# Open communication and browsers
agenticcli> slack
agenticcli> chrome

# Open text editors
agenticcli> notepad++

All application commands execute immediately without confirmation prompts, making it quick to launch your tools.

Exiting

Exit the CLI with a friendly agent-themed message:

agenticcli> /exit

Agent signing off. Goodbye! 😊

You can also press Ctrl+C to exit anytime, which shows the same goodbye message.

Configuration

Settings are saved in ~/.agenticcli-config.json

Available Settings

defaultModel - gemini or bedrock (default AI model)
temperature - 0.0 to 1.0 (AI creativity level, default: 0.7)
maxTokens - Maximum response length (default: 2048)
colorEnabled - Enable/disable colors (default: true)
autoExecuteSafe - Auto-execute safe commands without confirmation (default: false)

Managing Configuration

# View current settings
agenticcli> /config

Current Configuration:

  defaultModel: gemini
  temperature: 0.7
  maxTokens: 2048
  colorEnabled: true
  autoExecuteSafe: false

# Update a setting
agenticcli> /set temperature 0.9
Success: Set temperature = 0.9

agenticcli> /set defaultModel bedrock
Success: Set defaultModel = bedrock

# Reset to defaults
agenticcli> /reset
Reset all settings to defaults? (y/n): y
Success: Configuration reset to defaults

Color Scheme

AgenticCLI uses a carefully designed color scheme for better readability:

Yellow - Welcome banner ASCII art and box borders
Green - Status indicators (● Online, ● Ready, ● Connected)
Green - "Agent is thinking..." message
Red - Error indicators (● Error, ● No Keys)
Cyan - Active model display
White - Standard text and command output

Smart Input Detection

AgenticCLI automatically detects what you're trying to do:

Commands starting with / - Treated as CLI commands
Common system commands - Executed as shell commands (ls, pwd, cd, cat, etc.)
Questions and natural language - Sent to AI for processing

Examples:

agenticcli> /help          # CLI command
agenticcli> ls -la         # System command
agenticcli> what is git?   # AI query

Tips & Tricks

Quick Command Access: Type / and the first letter of a command, then press Tab
Launch Applications Fast: Just type the app name - excel, chrome, code, etc.
Context Awareness: Load files with /read before asking questions about them
History Management: Use /clear-history to start fresh conversations
Model Switching: Try different models for different tasks (Gemini for general, Bedrock for specific use cases)
Save Important Chats: Use /save-chat to export valuable conversations
Arrow Navigation: Use ↑/↓ to quickly select from filtered commands

Troubleshooting

Model Status Shows "● Error"

Check your API keys in .env file
Verify internet connection
Ensure API keys have proper permissions

API Status Shows "● No Keys"

Create a .env file from .env.example
Add at least one set of API keys (Gemini or Bedrock)

Commands Not Working

Make sure commands start with /
Use /help to see all available commands
Check for typos in command names

Version History

V2.0 Beta (November 2, 2025)

New Features:

Application Launcher - Launch applications directly from AgenticCLI without confirmation prompts
Software Commands - Added support for opening:
- Microsoft Excel (excel)
- Microsoft Word (word)
- Slack (slack)
- Chrome Browser (chrome)
- Visual Studio Code (code)
- Kiro IDE (kiro)
- Notepad++ (notepad++)
- Git Bash (git-bash)
Enhanced Help Menu - Added dedicated "Software Commands" section in /help
UI Update - Changed welcome banner color from yellow to orange

Improvements:

Application commands are recognized as system commands (not AI queries)
All software launch commands execute immediately without confirmation
Better command mapping for Windows platform

V1.0 Beta (Initial Release)

Multi-model AI support (Gemini 2.5 Flash Lite, AWS Bedrock Nova)
Real-time command filtering and autocomplete
Arrow key navigation for commands
Smart input detection (CLI commands, system commands, AI queries)
Dynamic status indicators
Conversation history management
File operations (read, analyze, explain)
Configurable preferences
Cross-platform support (Windows & Linux)

License

MIT

Name		Name	Last commit message	Last commit date
screenshots		screenshots
src		src
.env.example		.env.example
.gitignore		.gitignore
LICENSE		LICENSE
README.md		README.md
package-lock.json		package-lock.json
package.json		package.json

Latest commit

History

Folders and files

Repository files navigation