twilio-voice-conversation-relay

Pass

Audited by Gen Agent Trust Hub on Sep 14, 2026

Risk Level: SAFEINDIRECT_PROMPT_INJECTION
Full Analysis
  • Indirect Prompt Injection Surface: The skill provides templates for processing real-time speech transcripts via WebSockets and sending them directly to an LLM interface.
  • Ingestion points: In SKILL.md, user speech data is extracted from the voicePrompt field of WebSocket events.
  • Boundary markers: Code templates do not explicitly show boundary encapsulation or instruction delimiters, although the documentation text provides appropriate warnings instructing developers to implement them.
  • Capability inventory: No unexpected backend capabilities or file system operations are exposed in the examples.
  • Sanitization: No data filtering or escaping is implemented in the introductory quickstart snippets, making it a point of consideration for developers before deploying to production.
Audit Metadata
Risk Level
SAFE
Analyzed
Sep 14, 2026, 08:05 PM
Security Audit — agent-trust-hub — twilio-voice-conversation-relay