The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill processes external data files such as methylation matrices and BED files, which represents a theoretical surface for indirect prompt injection if data files contain malicious instructions. This is a inherent property of the skill's intended purpose for data processing.
Ingestion points: Data loading functions in SKILL.md such as load_methylation_data, load_bed_file, and load_probe_annotation read data from the local file system.
Boundary markers: The skill instructions do not specify the use of delimiters or boundary markers to distinguish between data content and potential instructions.
Capability inventory: The agent uses data to drive scientific calculations and fetch annotations through the tooluniverse library.
Sanitization: The skill performs format-specific parsing but does not include semantic sanitization of the data content.
[EXTERNAL_DOWNLOADS]: The skill utilizes the tooluniverse library to fetch biological and regulatory annotations from external sources including Ensembl, SCREEN, ENCODE, and ChIPAtlas.

tooluniverse-epigenomics