The Agent Skills Directory

[Indirect Prompt Injection] (HIGH): The skill demonstrates an attack surface where untrusted external content (genomic data files and user-defined coordinates) is processed and passed to execution-capable tools.
Ingestion points: Processes genomic data files including .bw, .bed, .gtf, and .cool. It also accepts string-based 'region' parameters.
Boundary markers: Absent. There are no delimiters or instructions to ignore embedded commands in the processed data.
Capability inventory: Executes pyGenomeTracks and igv via shell commands and Python's subprocess.run. It also writes configuration files (tracks.ini, igv_batch.txt) to the local filesystem.
Sanitization: No sanitization or validation of input variables (like $region) is shown, which could lead to shell injection if the agent interpolates user-provided text into the bash loops.
[Unverifiable Dependencies] (MEDIUM): The skill relies on external software packages (pygenometracks, IGV, and R libraries like Gviz) which must be present in the environment. These tools often have complex dependency trees and may execute with the same privileges as the agent.
[Dynamic Execution] (MEDIUM): The skill generates executable batch scripts for IGV (igv_batch.txt) and configuration files at runtime before passing them to interpreters/CLI tools.

bio-data-visualization-genome-browser-tracks