The Agent Skills Directory

[COMMAND_EXECUTION]: The skill programmatically invokes external bioinformatics binaries (bwa, samtools, gatk) using list-based arguments with the subprocess module, which prevents shell-related vulnerabilities.
[DYNAMIC_EXECUTION]: Included test scripts verify the primary analysis logic and the effectiveness of security controls by executing the skill's Python components in a controlled manner.
[DATA_EXFILTRATION]: The skill processes genomic data files and queries well-known bioinformatics services (such as MyVariant.info and Ensembl) to retrieve variant annotations. These network operations are limited to domain-specific metadata retrieval.
[INDIRECT_PROMPT_INJECTION]: The skill handles complex data formats like VCF and Excel, which represents an indirect prompt injection surface. This is a standard characteristic of data processing skills and is managed by core platform safety mechanisms.
[SAFE]: Primary scripts include a reject_canonical path validator that proactively blocks attempts to write or modify files within protected input data directories, enforcing strict workspace isolation.

tooluniverse-variant-analysis