image-generation
Fail
Audited by Gen Agent Trust Hub on Oct 16, 2026
Risk Level: CRITICALINDIRECT_PROMPT_INJECTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill documentation and examples describe processing unvalidated user-provided text (prompts) to generate visual content and execute CLI commands.
- Ingestion points: User-provided strings flow into the
promptargument across multiple TypeScript functions (generateImage,generateImageWithSize,generateImageBatch) and thez-aiCLI tool's--promptflag inSKILL.md. - Boundary markers: The instructions do not define clear delimiters (e.g., XML tags or triple backticks) to wrap user input or provide specific "ignore embedded instructions" directives to the agent.
- Capability inventory: The skill uses
fs.writeFileSync(file system write) and executes shell commands via thez-aiCLI tool (command execution). - Sanitization: There is no evidence of input validation, escaping, or filtering of the user-provided prompt before it is interpolated into system commands or SDK calls.
- [COMMAND_EXECUTION]: The skill provides numerous examples and instructions for using the
z-aiCLI tool. - Evidence: Examples in
SKILL.mddemonstrate executingz-ai image --prompt "..."commands. While documented for the primary purpose of the skill, this represents an execution capability that requires careful handling of user input to prevent shell-based attacks.
Recommendations
- CRITICAL: 1 file(s) identified as malware by FileRep - DO NOT USE
Audit Metadata