spark-troubleshooting
Pass
Audited by Gen Agent Trust Hub on Jul 27, 2026
Risk Level: SAFENO_CODE
Full Analysis
- [SAFE]: The skill consists strictly of markdown instructions and metadata. It does not include any executable code, scripts, or binary files.\n- [PROMPT_INJECTION]: The skill possesses a surface for indirect prompt injection because it processes user-provided telemetry and error logs.\n
- Ingestion points: User-supplied Spark cluster identifiers, symptoms, and job error messages.\n
- Boundary markers: None present in the skill instructions.\n
- Capability inventory: The skill uses AWS tools (sagemaker-unified-studio-mcp-troubleshooting and sagemaker-unified-studio-mcp-code-rec) to perform failure analysis and provide code recommendations.\n
- Sanitization: No explicit sanitization or filtering of the processed logs is specified.\n
- Contextual Note: This ingestion is a standard requirement for Spark troubleshooting. The risk is minimized by the use of official tools for a specific technical purpose.\n- [DATA_EXFILTRATION]: The skill requests technical job identifiers (Cluster ID, Job ID, etc.) and error messages. These are standard diagnostic parameters for AWS services and do not involve the unauthorized access or exfiltration of sensitive files or system credentials.
Audit Metadata