The Agent Skills Directory

[PROMPT_INJECTION]: The skill possesses an indirect prompt injection surface because it ingests untrusted user data—specifically podcast scripts and character images—which are then passed to downstream AI services (text-to-speech, image generation, and video animation).
Ingestion points: User-provided URLs in the phota/train command and script text in the inworld/text-to-speech-2 command.
Boundary markers: No explicit delimiters or instructions are provided to the agent to ignore potentially malicious content embedded within the user-provided scripts or image metadata.
Capability inventory: The skill utilizes the Bash tool to execute multiple infsh app run commands for training, image generation, audio synthesis, video rendering, and media merging.
Sanitization: There are no documented steps for sanitizing or validating the input text or image sources before they are processed by the automated pipeline.
[COMMAND_EXECUTION]: The skill relies on the Bash tool to interact with the platform's CLI (infsh). While the documented instructions use structured JSON for inputs, the presence of a shell environment combined with user-controlled parameters (like URLs and script text) necessitates careful handling to prevent potential command injection if input is not properly escaped by the underlying platform.

ai-podcast