debug-r-env
Pass
Audited by Gen Agent Trust Hub on Sep 4, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill generates and executes commands to fetch R packages from well-known registries including CRAN and Bioconductor. It also supports downloading and installing packages directly from GitHub repositories using
remotes::install_githubanddevtools::install_github.- [COMMAND_EXECUTION]: The skill executes local R environment commands to perform diagnostics, includingversion,installed.packages(),sessionInfo(), and.libPaths(). It also generates commands for package installation and management.- [INDIRECT_PROMPT_INJECTION]: The skill processes external content by scanning R scripts for package declarations, which creates a potential surface for indirect prompt injection. - Ingestion points: Reads local R scripts from the path provided in the
script_pathparameter. - Boundary markers: Absent; there are no specific delimiters to distinguish script content from agent instructions.
- Capability inventory: The skill can execute package installations and R diagnostic commands.
- Sanitization: The skill does not perform sanitization on the contents of the scripts it scans.
Audit Metadata