The Agent Skills Directory

[SAFE]: The skill implements a robust path validation mechanism in scripts/utils.R using the is_path_within and resolve_output_dir functions. This ensures that all file writes and reads for output artifacts and data records are restricted to the skill's root directory, effectively preventing path traversal attacks.
[SAFE]: The skill includes a resource management feature in scripts/utils.R via the enable_timeout function, which uses R's setTimeLimit to enforce a user-defined execution duration. This prevents potential denial-of-service scenarios from long-running bioinformatics computations.
[DYNAMIC_EXECUTION]: In scripts/io.R, the load_gsva_result function uses readRDS to deserialize saved analysis objects for visualization. While deserialization is a potential attack vector, the risk is mitigated by the skill's path enforcement, which ensures the tool only loads files from its own validated output directories.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied CSV/TSV data for expression matrices and group definitions. In scripts/io.R, it implements sanitization by coercing expression data to numeric types and checking for missing values, reducing the risk of processing malformed or malicious data inputs.

immune-pathway-analysis