workflow-optimize-credits
Pass
Audited by Gen Agent Trust Hub on Jul 15, 2026
Risk Level: SAFEPROMPT_INJECTION
Full Analysis
- [PROMPT_INJECTION]: Indirect prompt injection vulnerability surface identified. Ingestion points: Workflow node details are retrieved from the environment using the 'read' tool. Boundary markers: There are no explicit instructions to use delimiters or ignore instructions found within the processed workflow nodes. Capability inventory: The skill has the ability to modify workflows using 'edit_node' and perform tasks via 'execute_clay_action'. Sanitization: No validation or sanitization procedures are defined for the ingested workflow data before it is processed by the agent.
Audit Metadata