tooluniverse-immune-repertoire-analysis
Pass
Audited by Gen Agent Trust Hub on Oct 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill processes external immune repertoire data files which could contain malicious content.\n
- Ingestion points: The
load_airr_datafunction inSKILL.mdreads data from files likeclonotypes.txtandfiltered_contig_annotations.csv.\n - Boundary markers: None. The instructions do not define delimiters or warnings to ignore embedded instructions in the sequencing data.\n
- Capability inventory: The skill uses
pandasfor file reading,matplotlibfor file writing (rarefaction_curve.png), and thetooluniversepackage for network-based searches (IEDB_search_tcells,PubMed_search).\n - Sanitization: CDR3 sequences are interpolated directly into PubMed search queries (e.g., in
predict_specificity_vdjdb) without escaping or validation, creating a potential path where malicious data in the source file could manipulate downstream tool behavior.
Audit Metadata