"Top Python Libraries" Publication 400 Subscriptions 20% Discount Offer Link.
Cradle is an open-source multimodal AI Agent framework developed by the BAAI-Agents team, designed for General Computer Control (GCC). It enables large multimodal models to interact with various software and games like a human, using screenshot inputs and keyboard/mouse outputs.
General Objectives:
Supports any local software (e.g., games, Office, image/video editing tools)
Multimodal input: Takes screenshots as input and supports keyboard/mouse operation output
Autonomous capabilities: Built-in “cognitive reflection + skill updating” module for continuous self-optimization
Modular design: Balances high controllability and scalability, easily adaptable to new environments