fujitsu-mainframe
Pass
Audited by Gen Agent Trust Hub on Sep 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill is configured to ingest and analyze legacy source files from the local filesystem, which serves as a potential vector for indirect prompt injection.
- Ingestion points: Step 1 in
SKILL.mdinstructs the agent to search for and process multiple legacy file types including*.cbl,*.ncb,*.jcl,*.cpy, and*.sdf. - Boundary markers: The skill instructions lack specific delimiters or "ignore instructions" directives to prevent the agent from potentially following natural language commands hidden within COBOL comments or JCL documentation.
- Capability inventory: The skill includes the use of filesystem discovery tools (
find) and instructions for the generation of executable Java code, Kubernetes manifests, and web application components acrossSKILL.mdand thereferences/files. - Sanitization: No mechanisms for sanitizing the input source files or validating the extracted logic are described, which could allow malicious code patterns to be propagated into the generated modern code.
Audit Metadata