02-merge-patterns
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection as it processes data from Silver tables and metadata from YAML configuration files.
- Ingestion points: The template scripts (e.g.,
accumulating-snapshot-merge.py,periodic-snapshot-merge.py) read data usingspark.table()and configuration viayaml.safe_load(). - Boundary markers: No explicit delimiters or instructions are used to separate data from processing logic.
- Capability inventory: The skill performs
DeltaTable.merge()operations, which can alter production datasets in the Gold layer. - Sanitization: The implementation uses
yaml.safe_load()to prevent unsafe object instantiation during configuration parsing.
Audit Metadata