The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill retrieves biological and chemical data from reputable scientific institutions, including the European Bioinformatics Institute (EBI), the National Center for Biotechnology Information (NCBI), the U.S. Food and Drug Administration (FDA), and the Open Targets platform. These are well-known, trusted sources for research data.
[COMMAND_EXECUTION]: The skill uses curl to fetch JSON data from public APIs and processes the results using Python one-liners or provided scripts. The Python code is defined locally within the skill and does not execute remote scripts, making this pattern safe in this context.
[INDIRECT_PROMPT_INJECTION]: The skill ingests data from external biological databases which constitutes an attack surface. 1. Ingestion points: Data is retrieved from ChEMBL, PubChem, and OpenFDA APIs. 2. Boundary markers: The skill does not explicitly use boundary markers when presenting API results to the agent. 3. Capability inventory: The skill has capabilities for network access and command execution through the agent environment. 4. Sanitization: The scripts use standard JSON parsing, which is a safe way to handle structured data from external APIs.

drug-discovery