skills/legout/data-platform-agent-skills/data-engineering-storage-remote-access-integrations-delta-lake/Gen Agent Trust Hub
data-engineering-storage-remote-access-integrations-delta-lake
Pass
Audited by Gen Agent Trust Hub on Feb 17, 2026
Risk Level: SAFE
Full Analysis
- [Category 8] Indirect Prompt Injection (LOW): The skill establishes a data ingestion surface by reading external data from cloud storage into the agent's memory via Pandas and PyArrow DataFrames.
- Ingestion points:
DeltaTable.to_pandas()andDeltaTable.history().to_pyarrow_table()inSKILL.md. - Boundary markers: Absent; the skill does not implement delimiters or specific warnings to ignore instructions embedded within the Delta tables.
- Capability inventory: The skill utilizes
deltalake.write_deltalakefor file system writes and performs network operations to S3, GCS, and Azure endpoints. - Sanitization: Absent; data is loaded directly from the storage provider into memory without content validation or escaping.
- [Category 2] Data Exposure & Exfiltration (SAFE): While the skill provides templates for cloud authentication, it uses standard placeholders (e.g.,
AKIA...) and explicitly includes documentation advising users to use environment variables instead of hardcoding credentials. - [Category 4] Unverifiable Dependencies & Remote Code Execution (SAFE): The skill relies on well-known, industry-standard libraries (
deltalake,pyarrow) from the official PyPI registry.
Audit Metadata