skills/databricks-solutions/vibe-coding-workshop-template/multi-agent-genie-orchestration/Gen Agent Trust Hub
multi-agent-genie-orchestration
Pass
Audited by Gen Agent Trust Hub on Mar 8, 2026
Risk Level: SAFE
Full Analysis
- [PROMPT_INJECTION]: No malicious prompt injection or safety bypass instructions were detected. The patterns focus on structured intent classification and routing.\n- [CREDENTIALS_UNSAFE]: No hardcoded credentials or secrets were found. Configuration templates use clear placeholders for Space IDs (e.g., 'YOUR_BILLING_SPACE_ID').\n- [EXTERNAL_DOWNLOADS]: No external scripts or unverifiable packages are downloaded. Dependencies are limited to standard, well-known libraries like 'databricks-sdk' and 'langgraph'.\n- [DATA_EXFILTRATION]: No suspicious network activity or sensitive file access was identified. The skill utilizes official Databricks SDKs for all workspace interactions.\n- [SAFE]: The skill explicitly addresses data integrity via the 'NO LLM Fallback' pattern, which mandates failing with an error instead of allowing the LLM to fabricate data during query failures, protecting against hallucinations.
Audit Metadata