The Agent Skills Directory

[EXTERNAL_DOWNLOADS] (LOW): The skill recommends installing the polars library using uv pip install polars. While Polars is a reputable and widely-used open-source project, installing unpinned packages can introduce minor supply-chain risks.
[DATA_EXPOSURE & EXFILTRATION] (SAFE): The provided examples demonstrate standard file I/O operations (e.g., read_csv, write_parquet) for data analysis. There are no instances of hardcoded credentials, access to sensitive system paths (like ~/.ssh), or unauthorized data transmission to external domains.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill's primary function is to process external data, which constitutes a surface for indirect prompt injection. If an agent uses these tools to process data from an untrusted source that contains embedded instructions, the agent might inadvertently follow those instructions.
Ingestion points: pl.read_csv, pl.scan_csv, pl.read_parquet, and pl.read_json identified in SKILL.md and references/core_concepts.md.
Boundary markers: Not present in the example snippets.
Capability inventory: The skill enables file writing (write_csv, write_parquet, etc.) and mentions connectivity to cloud storage and databases.
Sanitization: No specific sanitization or validation logic for ingested data is demonstrated in the documentation.
[PROMPT_INJECTION] (SAFE): No patterns of direct prompt injection, such as instructions to ignore previous rules or bypass safety filters, were detected in the markdown files or code blocks.

polars