data-catalog
Pass
Audited by Gen Agent Trust Hub on Feb 27, 2026
Risk Level: SAFEPROMPT_INJECTIONNO_CODE
Full Analysis
- [PROMPT_INJECTION]: The skill documents API patterns that allow for the ingestion of potentially untrusted data from a Dataiku instance, creating an indirect prompt injection surface.
- Ingestion points: Retrieval of dataset metadata (label, description, tags) in
references/metadata-and-tags.mdand listing of data collection objects inreferences/data-collections.md. - Boundary markers: The provided code snippets do not demonstrate the use of delimiters or instructions to ignore embedded commands within the retrieved text.
- Capability inventory: The skill includes patterns for high-privilege operations such as modifying dataset schemas, updating instance-level semantic meanings, and changing data collection permissions.
- Sanitization: No sanitization or validation of the retrieved metadata is shown in the examples before the data is processed by the agent.
- [NO_CODE]: The skill is composed exclusively of Markdown files providing documentation and code examples. It does not include standalone executable scripts or binary files.
Audit Metadata