The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill clones the i4h-workflows repository from https://github.com/isaac-for-healthcare/i4h-workflows. This repository is maintained by the Isaac for Healthcare team at NVIDIA, which matches the skill's authoring organization.
[REMOTE_CODE_EXECUTION]: Code from the cloned repository is executed using the uv run command as part of the core dataset conversion and inspection workflow.
[COMMAND_EXECUTION]: The skill utilizes several shell commands for repository management and dataset processing, including git clone, git rev-parse, cd, mkdir, and uv run.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data from HDF5 recordings and associated Scene/embodiment manifest files, creating a vulnerability surface for indirect prompt injection.
Ingestion points: Metadata including workflow details, scene descriptions, embodiment labels, and task instructions are read from external recordings and manifest files.
Boundary markers: Absent; the instructions do not define delimiters or provide warnings to the agent to ignore instructions embedded within the processed data.
Capability inventory: The skill allows the agent to execute shell commands via git and uv and perform local file system operations.
Sanitization: No evidence of input sanitization or strict schema validation for the textual content of the manifest files was found.

i4h-workflow-dataset-convert