ducklake
Pass
Audited by Gen Agent Trust Hub on Sep 25, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill allows the agent to process and manipulate data stored in external lakehouse tables and remote storage backends (S3, R2, GCS, Azure, and PostgreSQL). This creates a vulnerability surface where instructions embedded in the external data could attempt to influence the agent's actions.
- Ingestion points: Data retrieved from DuckLake tables through SQL queries and snapshot inspections (SKILL.md).
- Boundary markers: None provided in the documentation to distinguish between data and instructions.
- Capability inventory: SQL execution including data mutation (INSERT, UPDATE, DELETE, MERGE), schema modification (ALTER TABLE), and storage maintenance calls (ducklake_cleanup_old_files).
- Sanitization: No explicit sanitization or validation of the retrieved table content is described.
Audit Metadata