The Agent Skills Directory

[COMMAND_EXECUTION]: The skill executes shell commands via the Bash tool to run R-specific workflows.
Evidence: Uses Rscript -e to call devtools and roxygen2 functions, and runs R CMD check for package validation.
[INDIRECT_PROMPT_INJECTION]: The skill reads and triages data from external R package files, which could contain adversarial instructions intended to influence the agent's output.
Ingestion points: Reads the DESCRIPTION file for package metadata and scans source code files during the review phase.
Boundary markers: No specific delimiters or safety instructions are used to isolate package content from the agent's reasoning instructions.
Capability inventory: The skill has access to Bash, Read, Grep, and Agent tools to act on the package data.
Sanitization: No sanitization or filtering is applied to the package content before it is processed.
[METADATA_POISONING]: The author field in the YAML frontmatter is set to "Claude Code Academic Workflow," which is inconsistent with the expected vendor identification but does not appear to be malicious.

r-package-check