The Agent Skills Directory

Command Execution: The skill automates the execution of a specialized Python script (run_scrnaseq_post_count_qc.py) located within the system's plugin directory. This is used to perform the core processing of sequencing data bundles and is an expected part of the analysis workflow.
File System Access: The workflow reads biological data from user-specified directories and writes various artifacts to the local filesystem, including processed AnnData objects, quality control tables, and visualization reports. These operations are necessary for the skill's primary function of data transformation and reporting.
Indirect Prompt Injection Surface: The skill processes external metadata from files such as manifest.tsv and dataset_metadata.json. This ingestion point is standard for bioinformatics pipelines to track sample information and batch parameters, serving as the interface between raw data and the analysis environment.
External Tool Integration: The instructions reference the use of established bioinformatics packages like scDblFinder, scVI, and Scanpy. These tools are standard in the single-cell research community and are used here to ensure robust doublet detection and data integration.

scrna-seq-qc