The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests untrusted data that is then processed by a hosted synthesis service, creating a vulnerability surface where malicious instructions could be embedded in input text or external references.
Ingestion points: The skill accepts text, voice_description, and --audio sources (which can be local paths or HTTPS URLs) from the user or project registry.
Boundary markers: The instructions do not mandate the use of specific delimiters or 'ignore' instructions for the synthesized text content.
Capability inventory: The skill utilizes the postplus CLI to perform network requests, file system writes (--output), and asset downloads.
Sanitization: The prompt relies on the hosted service's internal content policy (postplus_cli_hosted_media_content_policy_blocked) rather than local input sanitization.
[COMMAND_EXECUTION]: The skill is primarily a wrapper for the postplus CLI tool, executing various commands including media create, media poll, doctor, update, and auth login. While these are vendor-provided tools, they represent a high level of shell interaction.
[EXTERNAL_DOWNLOADS]: The skill includes instructions to fetch and install software updates using postplus update and to download generated audio files from remote servers via postplus media-file download.

voice-batch-runner