voice-ai-agent

Pass

Audited by Gen Agent Trust Hub on May 18, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: No security issues detected.
  • [PROMPT_INJECTION]: The skill inherently contains a surface for indirect prompt injection as it is designed to process untrusted voice input from callers to perform tasks. Ingestion points: Call transcripts processed via the handle_call webhook. Boundary markers: None are defined in the provided code snippets to separate system instructions from caller input. Capability inventory: The agent can write to Google Calendar (calendar.events().insert) and send SMS notifications (send_sms). Sanitization: No input validation or filtering is present in the example implementations. Because this capability is required for the skill's primary purpose, it does not escalate the security verdict.
  • [SAFE]: Implementation examples follow security best practices by using placeholders like your-api-key for sensitive credentials.
  • [SAFE]: The skill refers to established and well-known technology services such as Bland AI, Vapi, and Google Voice.
Audit Metadata
Risk Level
SAFE
Analyzed
May 18, 2026, 11:21 PM
Security Audit — agent-trust-hub — voice-ai-agent