tooluniverse-chemical-safety
Pass
Audited by Gen Agent Trust Hub on Apr 16, 2026
Risk Level: SAFE
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill retrieves scientific data from well-known and trusted external services including the FDA, PubChem (NIH), DrugBank, and the Comparative Toxicogenomics Database (CTD). These operations are consistent with the skill's primary purpose of chemical risk assessment.
- [COMMAND_EXECUTION]: The skill uses a variety of specialized tools (e.g.,
ADMETAI_predict_toxicity,PubChem_get_CID_by_compound_name) to query databases. All command patterns are specific to bioinformatics and chemical analysis, with no evidence of arbitrary shell execution or system-level command injection. - [INDIRECT_PROMPT_INJECTION]: The skill ingests data from external API sources such as drug labels and scientific database entries. While this represents a surface for indirect prompt injection, the risk is categorized as low because the data originates from highly regulated and trusted scientific repositories, and the skill implements a strict reporting structure that separates source data from analytical synthesis.
Audit Metadata