The Agent Skills Directory

[DYNAMIC_EXECUTION]: The diagnostic script scripts/inspect_dxpy.py uses dynamic module loading to verify the local Python environment.
Evidence: The script utilizes importlib.import_module to inspect symbols and signatures of the dxpy library, with module names provided from internal hardcoded dictionaries.
Context: This is a diagnostic utility intended to ensure the correct SDK versions and symbols are available locally for the skill to operate correctly.
[INDIRECT_PROMPT_INJECTION]: The skill processes various external configuration and workflow definition files, which represents an attack surface for indirect prompt injection.
Ingestion points: The skill reads and processes user-provided dxapp.json manifests, inputs.json files, and scientific workflow definitions (WDL, CWL, Nextflow).
Boundary markers: There are no specific delimiters or instructions provided to the agent to disregard potential instructions embedded within the content of these files.
Capability inventory: The skill is capable of executing system commands via the dx CLI, making network requests to the DNAnexus platform, and managing genomics compute resources.
Sanitization: To address these risks, the skill includes explicit instructions to quote shell arguments and use array-based subprocess calls when handling project names, paths, and metadata.
[EXTERNAL_DOWNLOADS]: The skill references and facilitates the acquisition of a platform-specific compiler tool from a remote source.
Evidence: The documentation directs users to download the dxCompiler JAR file from the official DNAnexus GitHub organization for workflow compilation tasks.
Context: The download targets the official repository of the platform provider, and the process is a standard part of the documented scientific workflow lifecycle.

dnanexus-integration