The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill describes workflows for loading and processing external data, creating an entry point for untrusted content.\n
Ingestion points: The skill uses pd.read_csv and sns.load_dataset to ingest data from files and remote repositories.\n
Boundary markers: There are no instructions or examples demonstrating the use of markers to delimit data from prompt instructions.\n
Capability inventory: The skill has access to the Bash, Read, Write, and Edit tools, which could be misused if instructions were injected via data.\n
Sanitization: The provided examples lack logic for sanitizing or validating ingested data before it is utilized in the agent's context.\n- [EXTERNAL_DOWNLOADS]: The skill fetches software and datasets from reputable and well-known platforms.\n
Evidence: It uses uv pip install to download the seaborn package from PyPI and fetches example data from the official seaborn-data repository on GitHub.\n
Evidence: The citation instructions involve fetching metadata from arxiv.org, a well-known scientific repository.\n- [COMMAND_EXECUTION]: The skill utilizes shell commands for environment setup and management.\n
Evidence: uv pip install commands are provided for installing the Seaborn library and its statistical extensions.

seaborn