env-and-assets-bootstrap

The 'env-and-assets-bootstrap' skill is designed to assist in the setup process of AI repositories that require a specific environment and asset preparation before running commands. This skill is aimed at ensuring that the environment is conservative and well-structured, focusing on setting up dependencies, checkpoints, and dataset paths. It addresses the problem of needing an environment that is compatible with a given repository's reproduction target, ensuring that users can prepare their system for running commands without unnecessary complexity. This skill does not handle tasks such as repo scanning, full orchestration, or interpreting academic papers; it is strictly for setup before running code.

Key features of this skill include generating conservative environment setup notes, recommending candidate conda commands, planning asset paths, and providing hints about the sources of checkpoints and datasets. The skill also identifies unresolved dependencies or asset risks that may impede the setup process. It relies on specific scripts, such as 'bootstrap_env.py' and 'plan_setup.py', as well as supporting policies outlined in 'env-policy.md' and 'assets-policy.md'. These features make the skill useful for users who are working on reproducing research experiments or setting up environments in a well-defined manner. Additionally, it is designed to be used only when the environment and asset setup is a precondition for running the commands in the repository.

This skill is ideal for developers, researchers, and data scientists who are focused on reproducibility and setup tasks. It is most useful when the user is aiming to ensure a proper environment is created before executing any code, especially in contexts where checkpoints, datasets, and cache directories are involved. Users should apply it only when a clear, credible reproduction target has been identified in the repository and setup is required before code execution.