The Agent Skills Directory

[SAFE]: The skill provides instructions for a multi-step media pipeline using platform-specific command-line tools (infsh app run). All tool invocations are consistent with the skill's stated purpose of generating AI podcast content.
[DATA_EXPOSURE]: The skill involves the processing of user-provided images and text. This data ingestion is localized to the media generation tools and is necessary for the skill's primary function. No sensitive local file paths (e.g., credentials, SSH keys) are accessed.
[INDIRECT_PROMPT_INJECTION]: The skill acts as a surface for processing external scripts and video prompts.
Ingestion points: User-provided conversation scripts (Step 4) and video prompts (Step 6).
Boundary markers: The skill does not explicitly define delimiters or instructions for the agent to ignore embedded commands in the user scripts.
Capability inventory: The skill has access to Bash (via infsh app run), Read, and Write tools.
Sanitization: No explicit sanitization or filtering of the input text is documented.
[REMOTE_CODE_EXECUTION]: The skill uses external tools (e.g., google/gemini-3-pro-image-preview, inworld/text-to-speech-2) through a platform-managed app runner. These are treated as vendor-integrated services and do not constitute arbitrary remote code execution.

ai-podcast