The Agent Skills Directory

[SAFE]: The skill provides official documentation, CLI scripts, and boilerplate code for integrating with ElevenLabs services. All external references point to official vendor domains (elevenlabs.io) or well-known, trusted services like Stripe, Vercel, and GitHub.
[INDIRECT_PROMPT_INJECTION]: The skill acknowledges the attack surface created by processing untrusted data through RAG (Retrieval-Augmented Generation) and tool inputs. It mitigates these risks by providing a structured 6-component framework for system prompts that includes explicit 'Guardrails' and 'Goals', and by demonstrating the use of Zod for strict schema validation in its code examples.

elevenlabs-agents