The Agent Skills Directory

[COMMAND_EXECUTION]: The skill provides technical instructions and code examples for executing shell commands using Python's subprocess.check_call. This functionality is utilized for running bioinformatics tools like samtools, managing environment dependencies through pip, and executing commands within Docker containers (references/app-development.md, references/configuration.md).\n- [EXTERNAL_DOWNLOADS]: The skill documents the installation of the dxpy toolkit from standard package registries and details patterns for downloading system packages and Docker images as part of the DNAnexus application build and execution process (SKILL.md, references/configuration.md).\n- [INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest and process genomics data and metadata from the DNAnexus platform, which represents a potential surface for indirect injection of malicious instructions.\n
Ingestion points: External content enters the agent's context through dxpy.download_dxfile and metadata retrieval via dxpy.find_data_objects (references/app-development.md, references/data-operations.md).\n
Boundary markers: While the documentation lists input validation as a best practice, the provided code examples do not incorporate specific prompt boundary markers to delimit external content (references/app-development.md).\n
Capability inventory: The skill possesses significant capabilities including file system access, network connectivity, and shell command execution across various scripts (references/app-development.md, references/configuration.md).\n
Sanitization: General recommendations for input validation are provided, but the examples do not demonstrate explicit sanitization or escaping of external data before it is used in command execution strings (references/app-development.md).

dnanexus-integration