model-calibration-curve
Pass
Audited by Gen Agent Trust Hub on Sep 23, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [SAFE]: The skill performs survival analysis and calibration assessment using the well-known and trusted R packages rms and survival. All analysis logic is implemented via standard statistical functions with no evidence of malicious intent.
- [SAFE]: Data validation is performed in scripts/functions.R to ensure the clinical CSV contains adequate samples (>=30) and events (>=10) and follows expected numeric/binary encoding.
- [SAFE]: External dependencies are installed from the official CRAN cloud mirror (https://cloud.r-project.org), which is a trusted and well-known service for R package distribution.
- [DYNAMIC_EXECUTION]: The skill dynamically constructs R survival formulas in scripts/functions.R using string concatenation of user-provided feature names. This behavior is limited to statistical formula parsing and incorporates backtick escaping to prevent command injection.
- [INDIRECT_PROMPT_INJECTION]: The skill ingests clinical data from a user-provided CSV file, representing a potential indirect prompt injection surface. 1. Ingestion points: Clinical CSV data is read into R via read.csv in scripts/run_analysis.R. 2. Boundary markers: The skill does not implement delimiters or ignore-embedded-instruction warnings for the raw data content. 3. Capability inventory: The skill maintains capabilities for file writing (.qs, .xlsx, .pdf, .txt) and supports elapsed time limits for computations. 4. Sanitization: Inputs are strictly coerced to numeric and factor types during data preparation, effectively neutralizing non-numeric content.
Audit Metadata