The Agent Skills Directory

[PROMPT_INJECTION]: Indirect prompt injection surface detected through data ingestion.
Ingestion points: The scripts scripts/eda.py, scripts/clean.py, and scripts/engineer_features.py read user-provided dataset files. The eda.py script specifically prints the top rows and categorical values to the console.
Boundary markers: No delimiters or safety instructions are used in the script output to designate data as untrusted or to prevent the agent from following instructions embedded within dataset records.
Capability inventory: The skill allows shell command execution (restricted to the provided scripts via uv run) and has permissions for reading and writing local files.
Sanitization: There is no sanitization or escaping of the dataset content before it is printed to the agent's context; raw values are displayed as-is.
[EXTERNAL_DOWNLOADS]: The skill uses uv run to manage its environment, which involves downloading verified dependencies such as pandas and numpy from official package registries if they are not present.

data-prep