buildroid 1.2.0

Automation of Android Project Building with LLMs

BuilDroid

⚡ Clone, build, and generate debugging APKs for Android projects using LLM-powered automation.

BuilDroid is a Python package that leverages Large Language Models (LLMs) to automatically clone any Android project hosted on GitHub, configure it, and build the debugging .apk file. This enables faster evaluation, performance testing, reverse engineering, or security analysis of Android applications. The building process happens in an isolated Docker container.

🚀 Features

• 🔗 Clone any Android GitHub repository.
• ⚙️ Auto-configure Gradle build for debugging.
• 🤖 LLM-guided build troubleshooting and error recovery.
• 📦 Outputs ready-to-install debugging APK.
• 🧪 Supports workflows for performance evaluation and static/dynamic analysis.

📦 Installation

pip install buildroid

📦 Dev Container Setup

To setup in a VSCode Dev Container:

1. Install the Dev Containers extension.
2. Clone this repository.
3. Open the repository in VSCode, and it will prompt you to reopen in the dev container. Alternatively, use a command to open the current folder in a dev container.

✅ Requirements

• Python 3.10+
• Git installed and accessible from terminal
• OpenAI API key (or other LLM provider) for LLM access.

⚙️ LLM Configuration

builDroid uses an LLM backend for build assistance. To use it:

1. Obtain your API key from OpenAI or compatible provider.
2. Set your API key as a .env file:

API_KEY=<your-api-key-here>
BASE_URL=<your-base-url-here>
LLM_MODEL=<your-llm-model-here>

BASE_URL and LLM_MODEL are optional. If not provided, builDroid will use OpenAI's gpt-4.1-mini-2025-04-14. For example, if you put 'https://generativelanguage.googleapis.com/v1beta/' as your base url, builDroid will access Google AI's gemini-2.0-flash-lite. If you want to use other providers, you have to provide the base url and the LLM model in .env.

3. (Optional) builDroid's primary goal is to successfully execute ./gradlew assembleDebug. To change its goals, create a ai_settings.yaml file in the working directory. The example file is in the source code.

🖥️ Usage

CLI Usage

buildroid build https://github.com/user/project # Run on a single repository
buildroid build repos.txt # Run on a list of repositories from a file
buildroid build local_path --local # Run with a local repository

buildroid clean # Clean test results

Advanced Options for Builds

• -n, --num: Specify cycle limit (max. number of commands to execute)
• -c, --conv: Enable conversation mode (API works with conversation models)
• -k, --keep-container: Keep container after build (builDroid removes container by default)
• -l, --local: Build from a local repository (Provide local path instead of Github link)

Python Usage

import builDroid

source = "https://github.com/user/project"

builDroid.utils.api_token_setup() # This function will setup the environment variables from .env file

builDroid.process_repository(repo_source=source)
# args: 
# repo_source: str,
# cycle_limit: int = 40,
# conversation: bool = False,
# extract_project: bool = True,
# override_project: bool = False,
# keep_container: bool = False,
# user_retry: bool = False,
# local_path: bool = False,
# project_name: str = None

builDroid.utils.api_token_reset() # This function will reset the environment variables

🛠️ Troubleshooting

If the build fails, builDroid will attempt to:

1. Analyze the error output.
2. Query the LLM for common solutions.
3. Retry the build with suggested fixes.

❗ Note: Some complex/outdated builds may still require manual intervention.

🏗️ Roadmap

• Integration with emulator for automated APK testing

🤝 Contributing

Pull requests are welcome! For major changes, please open an issue first to discuss.

📜 License

MIT License. See LICENSE for details.

🙏 Acknowledgments

• OpenAI for LLM API
• ExecutionAgent