The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied gene and lncRNA lists which represent a surface for potential indirect prompt injection. This is effectively mitigated by strict validation against biological databases.
Ingestion points: The --target_genes and --target_lncrna parameters processed in scripts/run_analysis.R via parse_optional_list in scripts/utils.R.
Boundary markers: None; the skill expects biological identifiers.
Capability inventory: The skill has local file system read/write access (scripts/io.R) and generates PDF plots.
Sanitization: Data is sanitized by intersecting the user-provided lists with internal database columns in scripts/functions.R, ensuring only known biological identifiers are processed.
[DYNAMIC_EXECUTION]: The skill uses the R source() function in scripts/main.R to load modular components. These operations use fixed relative paths determined by the script's own directory, preventing the loading of arbitrary or remote code.

lncrna-regulatory-network-construction-analysis