The Agent Skills Directory

[COMMAND_EXECUTION]: The skill instructs the agent to execute several shell commands to manage environment life cycles. Evidence includes:
Interactive project scaffolding using uv run isaaclab --new.
Dependency management and environment validation using uv sync and uv run pytest.
Execution of Python scripts for listing environments and running agents (python scripts/list_envs.py, isaaclab random_agent).
[INDIRECT_PROMPT_INJECTION]: The skill is designed to ingest user-provided descriptions of robot tasks to generate simulation code and project scaffolding, which constitutes an indirect prompt injection surface.
Ingestion points: User queries for creating locomotion tasks, reaching tasks, or training configurations (referenced in evaluations.md).
Boundary markers: None identified. The skill does not explicitly instruct the agent to use delimiters or ignore instructions embedded within the user's task description.
Capability inventory: The skill has the capability to execute shell commands (uv run), run tests (pytest), and generate new project directories and files through the isaaclab CLI.
Sanitization: Not explicitly present. The skill relies on the logic of the isaaclab --new generator and the LLM's own filters for handling user-provided strings like project names and task family names.

isaaclab-building-environments