llm-cli

A simple command-line tool that uses AI (Claude, OpenAI, or Ollama) to suggest shell commands, generate code snippets, or explain programming concepts based on natural language descriptions.

Features

• Command suggestions: Get shell commands from natural language descriptions
• Code gen: Generate code snippets with the --code flag
• Explanations: Get brief explanations of commands/concepts with the --explain flag
• Multi-API support: Works with Anthropic Claude, OpenAI GPT models, and local Ollama models

Installation

Install Go, run make install.

Setup

Set one of the following environment variables:

export ANTHROPIC_API_KEY=your_claude_api_key
export OPENAI_API_KEY=your_openai_api_key
export OLLAMA_MODEL=your_ollama_model_name

The tool will automatically use whichever key or model is available (Claude takes priority if multiple are set).

Usage

Basic Commands

% llm search for files larger than 100MB
find . -type f -size +100M

% llm decrypt with gpg, unzip, filter for files larger than 10gb, sum the third column
gpg --decrypt archive.gpg | unzip -p - | find . -type f -size +10G -exec awk '{sum += $3} END {print sum}' {} +

Code Generation

% llm -c python to port scan 10.8.1.1/24
import socket
from concurrent.futures import ThreadPoolExecutor

def scan_port(ip, port):
    try:
        sock = socket.socket(socket.AF_INET, socket.SOCK_STREAM)
        sock.settimeout(1)
        result = sock.connect_ex((ip, port))
        sock.close()
        if result == 0:
            print(f"{ip}:{port} open")
    except:
        pass

def scan_host(host):
    ip = f"10.8.1.{host}"
    with ThreadPoolExecutor(max_workers=100) as executor:
        for port in range(1, 1025):
            executor.submit(scan_port, ip, port)

with ThreadPoolExecutor(max_workers=50) as executor:
    for i in range(1, 255):
        executor.submit(scan_host, i)

Explanations

% llm --explain what does grep -r do
grep -r performs a recursive search through directories...

% llm -x explain the find command
The find command searches for files and directories...

Options

• -c, --code: Code generation mode
• -x, --explain: Explanation mode
• -h, --help: Show help message
• -v, --version: Show version

Models Used

• Claude: claude-sonnet-4-20250514
• OpenAI: gpt-4o-mini
• Ollama: Any locally installed model (e.g., llama2, mistral, codellama)

License

MIT