cchain

Overview

cchain is a command line tool designed to execute a series of commands based on a configuration file. It supports retrying commands if they fail, with a specified number of attempts. Additionally, cchain can generate command inputs using a language model (LLM) based on specified functions.

Features

• Execute commands with specified arguments.
• Retry commands on failure with configurable retry limits.
• Simple configuration using JSON files.
• Logging of command execution and retries.
• Generate command inputs dynamically using LLM functions.

Installation

Cargo

Use Cargo to install cchain:

cargo install cchain

Building from Source

To install cchain, clone the repository and build it using Cargo:

git clone https://github.com/aspadax/cchain.git
cd cchain
cargo build --release

Usage

Create a JSON configuration file with the commands you want to execute. Example configuration:

[
    {
        "command": "echo",
        "arguments": ["Hello, world!"],
        "retry": 3
    },
    {
        "command": "ls",
        "arguments": ["-la"],
        "retry": 1
    }
]

Additionally, if you do not specify a configuration file, cchain will list all available configuration files in the current working directory that start with cchain_ and have a .json extension. You can then select the desired configuration file by entering the corresponding number.

Example:

cchain

This will prompt you to select from the available configuration files in the current directory.

Run cchain with the path to your configuration file:

cchain --configuration_file path/to/configuration.json

Also, if you would like to pick a command chain in a different folder than the current one, you can use the --directory flag:

cchain --configuration_files path/to/the/directory

To generate a template configuration file, use the --generate flag:

cchain --generate

Using Functions with LLM

You can specify functions in your configuration file that will generate command inputs dynamically using a language model. Example configuration with a function:

[
    {
        "command": "echo",
        "arguments": ["Hello, world!"],
        "retry": 3
    },
    {
        "command": "git",
        "arguments": ["commit", "-m", "llm_generate('generate a commit message', 'git --no-pager diff')"],
        "retry": 1
    }
]

In this example, the llm_generate function will use the specified arguments to generate a git commit message by prompting the LLM with git --no-pager diff.

You can configure the LLM by setting the following environment variables:

export CCHAIN_OPENAI_API_BASE="http://localhost:11434/v1"
export CCHAIN_OPENAI_API_KEY="test_api_key"
export CCHAIN_OPENAI_MODEL="mistral"

Here in the example, we are using a locally hosted Ollama model.

License

This project is licensed under the MIT License.