The Agent Skills Directory

[SAFE]: No security issues detected. The skill provides legitimate technical documentation for the Ray Data library, referencing official repositories and standard package registries.
[EXTERNAL_DOWNLOADS]: The skill references standard library installation via 'pip install' for well-known packages (ray, pyarrow, pandas) and provides examples for reading data from common cloud storage providers (AWS S3, Google Cloud Storage). These references are documented neutrally as expected features of the tool.
[INDIRECT_PROMPT_INJECTION]: The skill documents ingestion of untrusted data from external sources. 1. Ingestion points: Data is read from S3, GS, and local paths via functions like read_parquet and read_csv in SKILL.md. 2. Boundary markers: Not present in provided examples. 3. Capability inventory: The skill enables processing via mapping functions and writing to storage. 4. Sanitization: No explicit sanitization of ingested data content is shown.

ray-data