The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill instructs the agent to install several standard data science and machine learning libraries using the uv package manager: wandb, mlflow, optuna, matplotlib, seaborn, and pandas. These are well-known technology packages from established providers and official registries.
[PROMPT_INJECTION]: The skill exhibits a standard indirect prompt injection surface as it is designed to ingest and process user-provided descriptions for RL tasks, environments, and desired behaviors.
Ingestion points: The agent processes user descriptions of a [task], [undesired behavior], or [algorithm] as defined in the skill's triggers.
Boundary markers: There are no explicit delimiters or specific instructions provided to the agent to treat this user input as untrusted or to ignore any potential instructions embedded within those descriptions.
Capability inventory: The skill's instructions do not define dangerous capabilities such as direct shell command execution or arbitrary file writing; its primary function is design guidance and configuration management.
Sanitization: No explicit sanitization or validation of user-provided task descriptions is mentioned.

data-scientist