The Agent Skills Directory

[SAFE]: The skill contains no executable code or shell scripts. It consists entirely of natural language instructions for prompt formatting and content transformation.
[PROMPT_INJECTION]: No attempts to override system instructions, bypass safety filters, or extract system prompts were detected. The language is purely functional.
[DATA_EXPOSURE]: The skill does not access sensitive local files or environment variables. It does not contain any hardcoded credentials.
[EXTERNAL_DOWNLOADS]: No external URLs, scripts, or package installations are referenced.
[INDIRECT_PROMPT_INJECTION]: The skill processes potentially untrusted storyboard data.
Ingestion points: Storyboard description and atmosphere fields are ingested to create prompts.
Boundary markers: The skill enforces a strict output schema (timestamp-based lines) and explicitly forbids creating content not found in the source description.
Capability inventory: The skill uses update_storyboard to save generated text. It does not have network access or general file system write capabilities.
Sanitization: The instructions include strict constraints against hallucinations and abstraction to ensure the output stays within the context of the input data.
[DYNAMIC_CONTEXT_INJECTION]: No load-time shell execution patterns were found.

video-prompt