The Agent Skills Directory

[INDIRECT_PROMPT_INJECTION]: The skill ingests user-supplied text for voiceover generation, which is then processed by remote tools, creating a potential surface for indirect prompt injection.
Ingestion points: The text field within the generate_speech tool calls, as documented in SKILL.md and references/workflow-basics.md.
Boundary markers: The skill provides specific syntax for interjections and pauses (e.g., <#x#>) and instructs the agent to "preserve exact wording," which serves as a basic delimiter between data and instructions.
Capability inventory: The skill utilizes tools capable of network operations (API calls to app.creativeclaw.co), file management (importing and uploading media), and media assembly (merging audio and video tracks).
Sanitization: No explicit sanitization, filtering, or escaping of the user-provided script text is described before it is interpolated into tool parameters.
[EXTERNAL_DOWNLOADS]: The skill configuration defines a dependency on an external Model Context Protocol (MCP) tool.
Evidence: The agents/openai.yaml file specifies a dependency on the creative-claw tool hosted at https://app.creativeclaw.co/mcp/chatgpt. This is a vendor-owned resource matching the skill author's domain.

creativeclaw-minimax-speech