# ai-cli

**Repository Path**: chambersxdu/ai-cli

## Basic Information

- **Project Name**: ai-cli
- **Description**: 一个简单的命令行LLM工具
- **Primary Language**: Unknown
- **License**: Not specified
- **Default Branch**: master
- **Homepage**: None
- **GVP Project**: No

## Statistics

- **Stars**: 0
- **Forks**: 0
- **Created**: 2025-05-03
- **Last Updated**: 2025-05-05

## Categories & Tags

**Categories**: Uncategorized

**Tags**: None

## README

# Simple AI CLI Assistant

A simple command-line interface (CLI) tool to interact with Large Language Models (LLMs) using an OpenAI-compatible API directly from your terminal. Designed for personal use and quick setup across machines.


## Installation

**⚠️ Security Warning:** The following command downloads and executes a script directly from the internet. This method is convenient but bypasses standard security checks. **Only run this command if you trust the source URL (i.e., you control the script being downloaded).**

**Run the Installer:** Open your terminal and execute:

```bash
curl -fsSL https://gitee.com/chambersxdu/ai-cli/raw/master/install_ai.sh | bash
```

If you want to install it on Windows, please open the Powershell and execute:

```bash
irm https://gitee.com/chambersxdu/ai-cli/raw/master/install.ps1 | iex
```


**What the installer does:**

*   Checks for prerequisites.
*   Installs the `requests` Python library using `pip3 install --user`.
*   Downloads the `ai.py` script.
*   Installs it to `~/bin/ai`.
*   Makes it executable (`chmod +x`).
*   Checks if `~/bin` is in your system's `PATH` and provides instructions if it's not. If you see a warning about PATH, you'll need to add `export PATH="$HOME/bin:$PATH"` to your shell configuration file (e.g., `~/.bashrc` or `~/.zshrc`) and restart your shell or source the file.



## Configuration

1.  **First Run:** Run the `ai` command once without arguments:
    ```bash
    ai
    ```
    This will create the default configuration file at `~/.ai_cli_config` if it doesn't exist.

2.  **Edit Configuration:** Open the configuration file with your preferred text editor:
    ```bash
    nano ~/.ai_cli_config
    # or vim, code, etc.
    ```

3.  **Set Required Values:**
    *   `api_key`: **REQUIRED.** Replace `YOUR_API_KEY_HERE` with your actual LLM API key.

4.  **Customize Optional Values:**
    *   `base_url`: Change if you are not using the official OpenAI API (e.g., for a local model server like Ollama: `http://localhost:11434/v1`).
    *   `default_model`: Set your preferred default model name.
    *   `models`: Edit the comma-separated list of models available through your `base_url`.
    *   `system_prompt`: Customize the instructions given to the AI.
    *   `request_timeout`: Adjust the API request timeout in seconds if needed.

## Usage

*   **Basic Query (uses default model):**
    ```bash
    ai "How to find files modified yesterday?"
    ```

*   **Specify a Model:**
    ```bash
    ai -m gpt-4o "Translate 'hello world' to French"
    ```

*   **List Configured Models & System Prompt:**
    ```bash
    ai -l
    ```

*   **Verbose Output (for debugging):**
    ```bash
    ai -v "Why is the sky blue?"
    ```


## ⚠️ Important Safety Warning

*   This tool sends your query to an external LLM API.
*   The AI model can generate **incorrect, inefficient, or dangerously destructive commands** (e.g., involving `rm -rf`, `sudo`, `chmod`, `dd`).
*   The system prompt asks the AI to be alert to dangerous commands, but based on testing, this is **NOT RELIABLE**. The AI may fail to identify risks or may not provide adequate warnings.
*   **YOU ARE RESPONSIBLE** for understanding and verifying any command suggested by this tool **BEFORE** executing it in your terminal.
*   **NEVER blindly execute commands generated by the AI, especially those involving `sudo` or potentially destructive operations.**