moda-image
Pass
Audited by Gen Agent Trust Hub on Oct 24, 2026
Risk Level: SAFEINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTIONCOMMAND_EXECUTION
Full Analysis
- [INDIRECT_PROMPT_INJECTION]: The skill facilitates the translation of user-supplied instructions into prompts for image generation models. This surface is managed by passing inputs as quoted arguments to platform-specific CLI tools.
- Ingestion points: User input for image generation prompts and transform instructions (SKILL.md).
- Boundary markers: Prompts are encapsulated within shell command arguments.
- Capability inventory: Shell command execution via the 'moda' CLI for media operations and file management.
- Sanitization: Instructions specify that prompts are relayed verbatim to the backend models.
- [DYNAMIC_EXECUTION]: The skill documentation details the use of a sandboxed JavaScript environment (moda canvas edit) and an XML-based markup engine (moda canvas markup). These are platform-native tools designed with security constraints, such as synchronous execution budgets, execution timeouts, and the blocking of sensitive APIs like network requests or arbitrary evaluation.
- [COMMAND_EXECUTION]: The skill utilizes a set of scoped and authorized CLI tools (moda:*) for interacting with design assets and media services. These operations are restricted to the platform's namespace and governed by the skill's allowed-tools configuration.
Audit Metadata