tooluniverse-infectious-disease
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFECOMMAND_EXECUTIONDYNAMIC_EXECUTIONEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [COMMAND_EXECUTION]: The
SKILL.mdfile contains instructions for the agent to "write and run Python code via Bash" when computation is required for statistics, data processing, or scoring. - [DYNAMIC_EXECUTION]: The skill directs the agent to generate Python scripts at runtime—using libraries such as
pandas,scipy,statsmodels, andmatplotlib—to analyze data retrieved from various ToolUniverse tools and report the results. - [EXTERNAL_DOWNLOADS]: The skill interacts with numerous external scientific services to retrieve genomic, proteomic, and pharmacological data. These include the NCBI Taxonomy and Datasets, UniProt, ChEMBL, DrugBank, KEGG, and Reactome. It also utilizes NVIDIA NIM (AlphaFold2, DiffDock) for protein structure prediction and docking, and searches literature via PubMed, EuropePMC (preprints), and OpenAlex. These are well-known services integrated for the skill's stated purpose of outbreak intelligence.
- [INDIRECT_PROMPT_INJECTION]:
- Ingestion points: The skill ingests untrusted data from external sources such as PubMed articles, EuropePMC preprints, and general web search results via
web_search. - Boundary markers: There are no specific instructions or delimiters provided to ensure the agent ignores potentially malicious instructions embedded within the titles, abstracts, or metadata of the retrieved scientific literature.
- Capability inventory: The agent has the capability to write files (Markdown reports and CSV data) and execute Python code through a shell interface.
- Sanitization: The instructions do not specify any sanitization or validation steps for the content retrieved from external APIs before it is processed or used to generate analysis scripts.
Audit Metadata