🚀 cchain

Automate like the future depends on it.
Replace brittle shell scripts with AI-powered, retry-aware workflows. Built in Rust for speed and reliability.

⚡ Quick Example

Automate Git commits with AI-generated messages (no more "fix typo" commits!):

cchain run ./cchain_git_commit.json # Using a pre-built workflow to commit changes with AI

JSON Workflow (cchain_git_commit.json):

[
  {
    "command": "git",
    "arguments": ["add", "--all"],
    "retry": 3
  },
  {
    "command": "git",
    "arguments": [
      "commit", "-m", 
      "llm_generate('Summarize these changes in 1 line', 'git diff --staged')"
    ],
    "failure_handling_options": {
      "exit_on_failure": true,
      "remedy_command_line": { "command": "git", "arguments": ["reset"] }
    }
  },
  {
    "command": "git",
    "arguments": ["push"],
    "retry": 2
  }
]

🌟 Why Developers Love cchain

🛠️ Features

🔗 Chain Commands Like a Pro

• Retry failed steps (up to N times, or use -1 to retry indefinitely until succeeded)
• Pass outputs between commands via environment variables
• Fix failures automatically: Roll back with remedy_command_line

🤖 AI-Powered Automation

"arguments": ["llm_generate('Summarize this error', 'cat crash.log')"]

• Integrate LLMs (OpenAI, local models via Ollama, etc.)
• Generate commit messages, error summaries, test data on the fly

🌐 Cross-Platform Consistency

• Works on Linux/macOS/Windows out of the box
• No more if [[ "$OSTYPE" == "linux-gnu"* ]]; then...

⚡ Performance You’ll Notice

• Built in Rust—starts faster than your shell’s && chain
• Uses 10x less memory than Python/Ruby scripts

📦 Installation

Cargo: You need to ensure that you have Rust installed on your system. Otherwise, you may need to follow the instructions here to install Rust: https://www.rust-lang.org/learn/get-started

If you are using a Debian system, you will need to have build-essentials installed. Below is an example for Ubuntu:

sudo apt update
sudo apt install build-essential

Install the tools.

cargo install cchain

Pre-built binaries You can navigate to the release section for downloading the latest binaries available.

🚀 Getting Started

1. Create Your First Workflow

cchain new deploy --prompt "Create a workflow to pull docker image from xxx, then run it in the background"

AI generates a starter cchain_deploy.json!

2. Run It!

cchain run ./cchain_deploy.json

3. Save for Later

cchain add ./cchain_deploy.json  # Bookmark it as workflow #0
cchain run 0  # Re-run anytime
cchain run deploy # Or, use keyword to run it
cchain run "deploy some other fancy stuff" # Or, use multiple keywords

4. Access Public Chains

You may also want to share your chain, or find chains created by someone else. I hosted a GitHub repository for this purpose:

git clone https://github.com/AspadaX/cchain-chains

This respository can be directly addded to your local bookmark:

cchain add https://github.com/AspadaX/cchain-chains

It is much welcomed to PR new chains to this repository!

🧩 Advanced Usage

Dynamic Environment Variables

{
  "command": "echo",
  "arguments": ["Building $APP_VERSION"],
  "environment_variables_override": {
    "APP_VERSION": "llm_generate('Generate a semantic version')"
  },
  "stdout_stored_to": "<<build_id>>"  # Pass to next command!
}

Concurrent Tasks (Beta)

[
  {
    "command": "xh download http://example.com/large-asset.zip",
    "concurrency_group": 1
  },
  {
    "command": "xh download http://example.com/large-asset.zip",
    "concurrency_group": 1
  },
  {
    "command": "xh download http://example.com/large-asset.zip",
    "concurrency_group": 1
  }
]  # Download 3 files in parallel

You may find examples in the ./examples directory of this repo. Also, you may use the following command to generate a template chain file:

cchain new your_file_name

🔍 Comparison

🛠️ Use Cases (Just mocks, but feasible)

CI/CD Made Simple

[
  { "command": "cargo test", "retry": 2 },
  { "command": "llm_generate('Write release notes', 'git log') > CHANGELOG.md" },
  { "command": "docker build -t myapp ." }
]

Developer Onboarding

cchain new setup --prompt "Clone repo, install deps, start services"

AI-Augmented Debugging

{
  "command": "llm_generate('Fix this error', './failing_script.sh 2>&1')",
  "stdout_stored_to": "<<fix_suggestion>>"
}

📚 Documentation

Guides

• Full JSON Schema
• LLM Configuration

🤝 Contributing

We welcome PRs!

📜 License

MIT © 2024 Xinyu Bao "Do whatever you want—just don’t make CI pipelines cry."