please do not escape

A dataset of sandboxes for AI coding agents, providing a curated list of runtime environments and policy wrappers to constrain agent behavior. It solves the problem of safely running AI coding agents by offering various isolation levels, network controls, and credential management options. The target audience includes individual developers, small teams, and platform builders working with AI coding agents. It is positioned as an open-source, community-contributed resource with a filterable interface, distinguishing itself through its comprehensive comparison of different sandboxing approaches.

Key features

• Filter by isolation type
• Filter by network mode
• Filter by credential handling
• Filter by open-source or product
• Filter by platform support
• Filter by agent compatibility
• View raw YAML data
• Contribute entries via PR