The Agent Skills Directory

[SAFE]: No malicious behavior or attack vectors were detected. The skill follows security best practices by providing an environment variable template for API keys rather than hardcoding credentials. Detailed analysis shows typical bioinformatics calculations performed on public scientific datasets.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface by retrieving genomic association data from external APIs (GWAS Catalog and Open Targets). This surface is mitigated by the provided Python code's use of strict regular expressions and numeric type casting when parsing association statistics, preventing arbitrary text injection from influencing the analysis logic.
[DYNAMIC_EXECUTION]: Instructions encourage the agent to generate and execute Python code for scientific computation. While this is a dynamic execution pattern, it is the primary intended function of the skill and is low-risk within the context of scientific data analysis with standard libraries like pandas and scipy.

tooluniverse-gwas-study-explorer