The Agent Skills Directory

[COMMAND_EXECUTION]: The skill includes scripts for environment verification and project scaffolding that execute system commands. Specifically, scripts/selftest.R uses system2 to run smoke tests of the scaffolding script and R CMD SHLIB to verify the C compiler on Windows. scripts/check_pandoc.py uses subprocess.run to verify the version of the Pandoc executable. These are standard procedures for validating the analysis environment.\n- [EXTERNAL_DOWNLOADS]: The scripts/selftest.R script fetches version information from https://cloud.r-project.org and https://cran.r-project.org to check for R and package updates. These are trusted official repositories for the R ecosystem.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes clinical data which constitutes an untrusted external input. It manages this risk through a phase-gated workflow that requires researcher approval at each step and by implementing data audits to identify inconsistencies before analysis.\n- [SAFE]: The skill implements best practices for clinical data security, including automated configuration of .gitignore to ensure patient-level data remains on the local system and is not committed to version control.

ophthal-analysis