nano-banana-2
Pass
Audited by Gen Agent Trust Hub on Mar 2, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSREMOTE_CODE_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill provides instructions to download and install the vendor's command-line interface (CLI) tool. This is a standard setup procedure for the service and targets the vendor's official domain.
- [REMOTE_CODE_EXECUTION]: The skill suggests an installation pattern using
curl -fsSL https://cli.inference.sh | sh. This executes a remote shell script from the vendor's server to facilitate the installation of theinfshtool. This behavior is consistent with the vendor's primary service delivery. - [COMMAND_EXECUTION]: The skill requires the
Bash(infsh *)tool to execute commands through the vendor's CLI. This is necessary for the skill to perform its intended image generation tasks. - [INDIRECT_PROMPT_INJECTION]: The skill accepts user-defined prompts and external image URLs as input for the image generation model. This creates a surface for potential indirect prompt injection if an attacker-controlled image or text is passed to the skill. There are no explicit delimiters or sanitization steps mentioned in the skill definition to mitigate this risk.
Audit Metadata