hyperframes-core
Pass
Audited by Gen Agent Trust Hub on Nov 14, 2026
Risk Level: SAFECOMMAND_EXECUTIONINDIRECT_PROMPT_INJECTIONDYNAMIC_EXECUTION
Full Analysis
- [COMMAND_EXECUTION]: The skill relies on an external CLI suite invoked through
npxfor core operations like linting and rendering. It also recommends using general shell execution tools as a fallback when platform-specific subagent delegation is restricted. - [DYNAMIC_EXECUTION]: For parallel task management, the skill instructs agents to generate prompt files and execute them as background shell processes ('Headless CLI children'), creating a complex runtime execution environment.
- [INDIRECT_PROMPT_INJECTION]: The core workflow ingests untrusted user content to drive automated generation tasks.
- Ingestion points: Data is read from project-relative
STORYBOARD.mdandSCRIPT.mdfiles. - Boundary markers: There are no documented instructions for wrapping or delimiting user content when it is interpolated into subagent dispatch packets.
- Capability inventory: User-provided narrative and metadata influence subagent prompts, CLI parameters, and TTS synthesis arguments.
- Sanitization: The manifest parser is explicitly lenient, preserving unknown keys and potentially facilitating the pass-through of hidden instructions to child agents.
Audit Metadata