ibm-mainframe
Pass
Audited by Gen Agent Trust Hub on Sep 14, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documentation establishes a workflow for the agent to ingest and analyze untrusted data from mainframe systems. 1. Ingestion points: Dataset inventory and SMF/RMF log extraction described in SKILL.md and mainframe-modernization.md. 2. Boundary markers: There are no instructions for using delimiters to isolate untrusted data from the agent's prompt instructions. 3. Capability inventory: The skill assumes capabilities for file system access, JCL job submission, and the use of CLI tools like Zowe and SonarQube. 4. Sanitization: No data validation or sanitization steps are defined for the extracted mainframe content.
Audit Metadata