Cradle is a Python framework that enables foundation models to control computer tasks via screenshots and keyboard/mouse actions, supporting multiple games and software. It provides environment configs, a modular structure, and instructions for setup and per-game usage.
Collecting history — the radar snapshots this repo daily. The trend line appears after 3 days of data (1 so far).
What it is
Cradle is a Python framework that empowers nascent foundation models to perform complex computer tasks via a unified interface that uses screenshots as input and keyboard/mouse operations as output. It targets general computer control across multiple games and software.
How it works
The repository organizes core modules for environment interaction, planning, memory, and providers. It includes a runner that coordinates execution across different games/software and a modular structure for environment configuration, skill execution, and LLM/CLI integrations. It references readiness for various games and software via per-game environment configurations and resource directories.
Getting started
Install and setup steps are provided as part of the README: "git clone https://github.com/BAAI-Agents/Cradle.git" (implied in setup) and subsequent environment preparation with Python 3.10. It includes creating a conda environment and installing requirements: "# Clone the repository\ngit clone https://github.com/BAAI-Agents/Cradle.git\ncd Cradle\n\n# Create a new conda environment\nconda create --name cradle-dev python=3.10\nconda activate cradle-dev\npip install -r requirements.txt" Additionally, there is an instruction to install OCR tools with spaCy models: "1. Option 1\n# Download best-matching version of specific model for your spaCy installation\npython -m spacy download en_core_web_lg" or an alternative path using a tar.gz/.whl URL.
The repository also points to per-game setup pages under docs/envs for Red Dead Redemption 2, Stardew Valley, Cities: Skylines, Dealer's Life 2, and Software.
Recent releases
The releases section indicates "latest 0: none"; no tagged releases are listed in the provided data.
Traction
The repository has 2559 stars, 267 forks, and 22 open issues. The last push was on 2024-11-07. There is no per-day star metric in the data provided, so only the total counts are cited here.
Behind the repo
Not applicable: there is no linked startup or company information in the provided FACTS data.
Caveats
License is MIT. The README includes environment setup, sample .env keys for OpenAI, Claude, and Azure, and keys for various services, which implies sensitive configuration; users must supply their own API keys. The latest release section shows none, indicating no named latest release in the provided data. The project appears Python 3.10 compatible per badge in README.