The Agent Skills Directory

[SAFE]: The skill provides legitimate instructions and code for single-cell ATAC-seq analysis, referencing peer-reviewed methodologies (Pliner et al. 2018, Bravo Gonzalez-Blas et al. 2023). The implementation follows standard R and Python practices for bioinformatics workflows.
[EXTERNAL_DOWNLOADS]: The skill includes instructions to install necessary software dependencies from reputable sources.
In usage-guide.md, it directs the user to install R packages from GitHub using remotes::install_github for repositories belonging to well-known academic labs: cole-trapnell-lab (University of Washington) and GreenleafLab (Stanford University).
It also provides instructions to install scenicplus via git clone from the aertslab repository (VIB-KU Leuven) followed by pip install.
These downloads are standard for the bioinformatics field and target the official repositories of the tools' developers.
[INDIRECT_PROMPT_INJECTION]: The skill possesses a data ingestion surface as it is designed to process user-supplied genomic datasets.
Ingestion points: In examples/cicero_workflow.R, the script reads peak_matrix, peak_metadata, and cell_metadata from file paths provided via command-line arguments.
Boundary markers: None explicitly defined in the provided code snippets.
Capability inventory: The R script has the capability to read local files and write results to the filesystem using write.table and write.csv.
Sanitization: The script relies on standard R parsers (readMM, read.delim) which interpret the data as numeric matrices or tab-separated text. The genomic data being processed (peaks and coordinates) does not provide a viable path for command injection or prompt manipulation in this context.

bio-atac-seq-co-accessibility