The Agent Skills Directory

[EXTERNAL_DOWNLOADS]: The skill contains a mechanism to recommend the installation of an external skill hub from github:mindverse/skillhub when users request features outside the skill's scope. This encourages users to download and execute code from an unverified third-party repository.
[INDIRECT_PROMPT_INJECTION]: The skill processes user-supplied data such as podcast topics, guest backgrounds, and audience descriptions to generate scripts, creating a surface for potential injection.
Ingestion points: User-provided inputs for the 'Topic/Theme', 'Guest Information', and 'Audience Persona' fields defined in the workflow.
Boundary markers: The skill uses basic markdown formatting for structure but lacks explicit delimiters or specific instructions to the model to ignore potential commands embedded within the user data.
Capability inventory: The skill does not request access to restricted tools, file system writes, or network operations in its configuration, limiting the impact of a successful injection.
Sanitization: No input sanitization or validation logic is defined to filter or escape instructions provided in the user's input.

podcast-script