skills/databricks-solutions/vibe-coding-workshop-template/lakebase-memory-patterns/Gen Agent Trust Hub
lakebase-memory-patterns
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSPROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill utilizes several official and well-known libraries consistent with the Databricks ecosystem, including databricks-sdk, databricks-langchain, and mlflow. These are trusted dependencies for building stateful applications on the platform.
- [PROMPT_INJECTION]: The skill implements a persistent memory architecture which represents an indirect prompt injection surface. Ingestion points: Data enters via user-specific memory storage (LongTermMemory) and session checkpoints (ShortTermMemory). Boundary markers: None identified in the provided implementation. Capability inventory: Provides tools for memory retrieval, storage, and deletion within Databricks Lakebase. Sanitization: User identifiers are sanitized for safe namespace isolation, and memory data is processed through standard JSON parsing. This is considered safe as it is the core intended functionality and uses governed storage.
- [SAFE]: No malicious patterns such as obfuscation, privilege escalation, or unauthorized data exfiltration were found. The skill follows best practices for secure state management, including token-based authentication and governed Unity Catalog storage.
Audit Metadata