The Agent Skills Directory

[COMMAND_EXECUTION]: The skill is designed to generate and execute analysis scripts across multiple languages (R, Python, Stata, Julia). These operations are restricted to specific tools and project-relative paths, minimizing the risk of arbitrary system command execution.- [EXTERNAL_DOWNLOADS]: The skill documentation refers to standard package management operations for established data science libraries, such as 'uv pip install' for Python and 'install.packages' for R. These references target official registries and well-known libraries.- [REMOTE_CODE_EXECUTION]: Analysis scripts are generated at runtime and executed to produce results. This process is protected by a multi-phase workflow that includes an automated code review by a separate agent and a requirement for a locked research design before estimation code is run.- [SAFE]: The skill follows security best practices for data science environments, including explicit rules against modifying raw data and requirements for standard script headers and reproducibility settings.

data-analysis