local-ai-infra-generation 1.1.0

A tool to generate infrastructure files (Dockerfile, etc.) for codebases using local LLMs.

local-AI-infra-generation

local-AI-infra-generation leverages local Large Language Models (LLMs) to analyze code repositories and automatically generate infrastructure files such as Dockerfiles and docker-compose.yml. All processing is performed locally, ensuring privacy and control over your codebase.

---

Features

• Codebase Embedding: Index and embed your codebase for semantic search and retrieval.
• Natural Language Q&A: Ask questions about your codebase and receive context-aware answers.
• Automated Infrastructure Generation: Generate Dockerfiles and docker-compose.yml files tailored to your project.
• Multi-language Support: Works with Python, JavaScript, TypeScript, and Go projects.

---

Getting Started

1. Prerequisites

• Python 3.11+
  Ensure you have Python 3.11 or higher installed.
  Check with:

  python --version
  

• Ollama
  Download and install Ollama for local LLM inference.

• C/C++ Build Tools
  Required for building tree-sitter-languages.

2. Installation

Create and activate a virtual environment:

python -m venv .venv
source .venv/bin/activate  # On Windows: .venv\Scripts\activate

Install the package from this repository:

pip install .

This will install the infra-gen command-line tool and all necessary dependencies.

---

Usage

1. Start Ollama

Make sure the Ollama server is running in the background:

ollama serve &

2. Run the CLI

Once installed, you can use the infra-gen command.

infra-gen --help

Common Commands

• Embed a Project:

  infra-gen embed /path/to/your/project
  

• Ask a Question:

  infra-gen ask "How does authentication work?" --project your_project_name
  

• List Embedded Projects:

  infra-gen list
  

• Generate Full Infrastructure (Dockerfile, Compose, etc.):

  infra-gen generate-infra /path/to/your/project --output ./infra
  

• Generate Only a Dockerfile:

  infra-gen generate-docker --project your_project_name
  

• Generate Only a docker-compose.yml:

  infra-gen generate-compose --project your_project_name
  

---

Configuration

The tool uses a config.yaml file for settings. The configuration is loaded in the following order of priority:

1. Via --config flag: Provide a direct path to a .yaml file.

   infra-gen --config /path/to/my-config.yaml embed /path/to/project
   

2. User-level config: Place a file at ~/.config/infra-generator/config.yaml.
3. Default package config: If no other config is found, a default version bundled with the package is used.

You can customize model names, ChromaDB storage directories, Ollama URLs, and more in your custom config file.

---

Development

If you want to contribute to the development of this tool, you can install it in editable mode.

1. Clone the repository:

   git clone https://github.com/yourusername/local-AI-infra-generation.git
   cd local-AI-infra-generation
   

2. Create and activate a virtual environment:

   python -m venv .venv
   source .venv/bin/activate
   

3. Install in editable mode:

   pip install -e .
   

This allows you to make changes to the source code and have them reflected immediately when you run the infra-gen command.

Running Tests

TODO: Add unit tests and instructions for running them.

---

Troubleshooting

• Ollama not found:
  Ensure Ollama is installed and available in your PATH.

• tree-sitter language .so files missing:
  If you encounter errors about missing .so files, ensure tree-sitter-languages is installed and built correctly.

• Model download issues:
  The first run will download required models. Ensure you have a stable internet connection.

---

TODO

• Add comprehensive unit and integration tests.
• Improve error handling and user feedback.
• Add support for more programming languages (e.g., Java, Rust).
• Enhance prompt templates for better infrastructure generation.
• Add web or GUI interface.
• Document API for programmatic usage.
• Support for private model registries and custom LLMs.
• Optimize embedding and retrieval for large codebases.
• Add CI/CD pipeline for automated testing and deployment.

---

License

This project is licensed under the Apache 2.0 License.