tooluniverse-small-molecule-discovery
Pass
Audited by Gen Agent Trust Hub on Sep 16, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSCOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill instructs the agent to install the
tooluniverse[ml]package viapipto enable machine-learning-based ADMET predictions. - [COMMAND_EXECUTION]: The skill encourages the agent to write and execute Python code using
bashfor data processing, statistical analysis (usingpandas,scipy,statsmodels), and visualization (matplotlib). - [INDIRECT_PROMPT_INJECTION]: The skill processes external data such as chemical names and SMILES strings from users or scientific databases which could serve as an attack surface for indirect prompt injection.
- Ingestion points: Parameters like
compound_name,smiles, andqueryin tools likePubChem_get_CID_by_compound_nameandChEMBL_search_molecules. - Boundary markers: Absent; there are no instructions to the agent to treat input identifiers as untrusted data or ignore embedded commands.
- Capability inventory: Shell command execution (Bash), Python execution, package management (
pip), and network access to scientific APIs. - Sanitization: Absent; no validation or escaping of SMILES strings or chemical names is described before they are passed to tools or included in analysis scripts.
- [DYNAMIC_EXECUTION]: The skill instructs the agent to generate and run Python scripts at runtime for data enrichment and scoring.
Audit Metadata