The Agent Skills Directory

[DATA_INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied CSV files, which presents a surface for indirect prompt injection if the agent interprets text content within the data dictionary as instructions.
Ingestion points: scripts/lint_data_dictionary.py (reads CSV data provided by the user).
Boundary markers: Absent; the skill does not explicitly use delimiters or instruction-ignore warnings when reading data files.
Capability inventory: The skill utilizes Read, Write, Bash, WebSearch, and WebFetch tools.
Sanitization: The linter validates column structure and data types but does not sanitize or escape natural language content in fields like 'Field Label' or 'Field Note'.
[EXTERNAL_DOWNLOADS]: The skill references external documentation from CDISC and NCI-EVS (National Cancer Institute).
Evidence: Links to cdisc.org for standards documentation are included in references/cdisc_mapping.md.
Analysis: These resources are from well-known professional organizations and are used solely for informational reference, posing no security risk.

alterlab-redcap-cdisc