The Agent Skills Directory

[COMMAND_EXECUTION]: The skill documentation provides instructions and examples for defining rules that execute shell commands using the shell: directive. This is the primary mechanism for Snakemake to automate external tools like bwa, samtools, and fastp.
[DYNAMIC_EXECUTION]: The documentation describes features for executing inline Python code via run: blocks and external R, Python, or Julia scripts via the script: directive. These capabilities allow for runtime script generation and execution based on workflow parameters.
[INDIRECT_PROMPT_INJECTION]: The skill outlines a workflow system that processes untrusted external data and configuration files.
Ingestion points: The agent processes user-provided Snakefile rules, config/config.yaml settings, and external data files (e.g., FASTQ, BAM).
Boundary markers: No explicit boundary markers or instructions to ignore embedded commands are demonstrated in the examples.
Capability inventory: The defined workflows include capabilities for file system operations, tool invocation, and network access (demonstrated via wget for downloading reference data).
Sanitization: The examples do not include sanitization or validation steps for input data or configuration values before they are interpolated into shell commands or script logic.

snakemake-workflow-engine