ingestion-pipeline-design
Pass
Audited by Gen Agent Trust Hub on Sep 12, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONEXTERNAL_DOWNLOADS
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied source requirements and data samples to generate architecture designs. It mitigates risks associated with untrusted data by instructing the agent to only request sanitized samples, explicitly label all inputs (Supplied, Documented, Inferred, Unknown), and strictly avoid generating executable commands or production configurations.
- [EXTERNAL_DOWNLOADS]: The skill uses the
webtool to access public documentation athelp.splunk.com. These downloads are restricted to technical reference materials for validation and decision-making within the design process and do not involve the execution of remote scripts. - [DATA_EXPOSURE]: The instructions include proactive safeguards against credential theft and data exposure. It explicitly directs the agent to never request or repeat secrets, API tokens, certificates, or private endpoints, mandating the use of placeholders for all sensitive parameters.
Audit Metadata