alterlab-redcap-cdisc
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [SAFE]: No security issues or malicious patterns were identified. The skill correctly limits its scope to offline design tasks and uses a secure, standard-library Python script for data validation.\n- [INDIRECT_PROMPT_INJECTION]: The skill processes user-provided CSV data to design or lint REDCap data dictionaries, which is its primary intended function.\n
- Ingestion points:
scripts/lint_data_dictionary.pyand the agent ingest variable names, labels, and branching logic from user-supplied CSV files.\n - Boundary markers: The skill relies on standard CSV formatting; no explicit delimiters or instructions to ignore embedded content are defined in the main prompt.\n
- Capability inventory: The skill is authorized to use
Bash,python,Read,Write,Edit,WebSearch, andWebFetch.\n - Sanitization: The
scripts/lint_data_dictionary.pyscript performs structural and syntax validation of REDCap expressions but does not filter for potential prompt injection markers.
Audit Metadata