The Agent Skills Directory

[SAFE]: The skill provides a controlled interface for voice matching, incorporating mandatory consent checks that require the agent to stop and confirm authorization before utilizing any reference audio.
[SAFE]: External media handling is restricted to the vendor's infrastructure through dedicated import tools, ensuring that audio assets are processed within the platform's security boundaries.
[INDIRECT_PROMPT_INJECTION]: The skill accepts user-provided script text and audio reference URLs to generate synthesized speech assets.
Ingestion points: User-supplied text and audio_url parameters are passed to the generate_speech tool.
Boundary markers: The instructions mandate preserving exact speech in the text field and using predefined expressive tags (<laugh>, <sigh>, etc.).
Capability inventory: The skill utilizes tools for audio generation, sequential merging, and film assembly, all scoped to the vendor's MCP implementation.
Sanitization: The skill documentation notes that the server-side process converts compatible bracketed reactions and removes incompatible delivery tags before generation.

creativeclaw-chatterbox