elevenlabs
Pass
Audited by Gen Agent Trust Hub on Oct 3, 2026
Risk Level: SAFEEXTERNAL_DOWNLOADSINDIRECT_PROMPT_INJECTION
Full Analysis
- [EXTERNAL_DOWNLOADS]: The skill includes a utility script
scripts/openapi_lookup.pythat fetches the official OpenAPI specification from ElevenLabs (https://api.elevenlabs.io/openapi.json). This is a well-known service, and the script uses standard Python libraries to fetch and cache the specification locally for lookup purposes. - [INDIRECT_PROMPT_INJECTION]: The skill provides instructions for building AI agents that ingest untrusted data from multiple sources, including audio transcripts (
scribe_v2), uploaded files in various formats (PDF, DOCX, etc.), and external website crawls. These agents are equipped with capabilities such as executing webhooks, connecting to MCP servers, and managing telephony transfers. The documentation mitigates these risks by providing detailed guidance on implementing safety guardrails, PII redaction, and strict security settings for tool execution. - [SAFE]: The skill promotes security best practices, specifically instructing that sensitive credentials like
xi-api-keymust remain server-side and never be exposed in client-side bundles. It details the process for minting short-lived, single-use tokens for browser-based operations.
Audit Metadata