polars
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for ingesting and processing data from various external sources including CSV, Parquet, JSON, Excel, databases, and web fetches from arXiv, which creates a surface for indirect prompt injection. The instructions do not define boundary markers or sanitization logic to protect the agent from instructions embedded in this untrusted data. * Ingestion points: Data reading functions such as read_csv, read_parquet, read_json, read_excel, and read_database in references/io_guide.md, as well as citation data fetching from arxiv.org in SKILL.md. * Boundary markers: None identified. * Capability inventory: The skill utilizes capabilities for file system writes, network communication for database/cloud storage, and general tool execution. * Sanitization: None identified.
- [SAFE]: The skill documentation follows security best practices by recommending managed identities, IAM roles, and specific credential providers over hardcoding secrets.
- [SAFE]: Installation instructions and external links target verified packages on official registries and well-known academic services.
Audit Metadata