The Agent Skills Directory

[COMMAND_EXECUTION]: The skill utilizes the Bash tool to execute shell commands, including Rscript with inline code execution (-e) to process data from local files. These commands incorporate file paths provided as user input to the skill.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from manuscripts and analysis output directories to extract numeric claims.
Ingestion points: The manuscript file ($0) and the outputs directory ($1) are the primary entry points for external data.
Boundary markers: Absent. The skill instructions do not specify any delimiters or ignore-instructions to protect against malicious content embedded in the manuscript or data files.
Capability inventory: The skill has broad capabilities, including Bash for command execution, Write for file creation, and Task for managing operations.
Sanitization: No explicit sanitization or input validation is mentioned for the data extracted from the manuscript before it is used for reporting.
[DATA_EXPOSURE]: The skill reads from potentially sensitive project directories, including the manuscript and all analysis outputs (RDS, CSV, logs, etc.), and generates reports in a quality_reports/ directory.

audit-reproducibility