vapi-calls

Pass

Audited by Gen Agent Trust Hub on Mar 31, 2026

Risk Level: SAFE
Full Analysis
  • [SAFE]: The skill's primary function is to facilitate legitimate voice communication via the Vapi API. It does not contain any malicious code, obfuscated instructions, or unauthorized data access patterns.
  • [EXTERNAL_DOWNLOADS]: The skill interacts with vapi.ai, which is a well-known technology service for voice AI. These network operations are documented and essential for the skill's functionality.
  • [CREDENTIALS_UNSAFE]: The documentation correctly identifies the need for a VAPI_API_KEY and explicitly instructs users to store it securely in the environment rather than hardcoding it, which aligns with standard security recommendations.
  • [PROMPT_INJECTION]: The skill has an attack surface for indirect prompt injection. 1. Ingestion points: Task-specific instructions are ingested into assistantOverrides in SKILL.md. 2. Boundary markers: No delimiters are specified to isolate untrusted instructions. 3. Capability inventory: The skill performs outbound network requests to the Vapi API (POST /call) as defined in SKILL.md. 4. Sanitization: No input validation or sanitization is described for the dynamic instructions.
Audit Metadata
Risk Level
SAFE
Analyzed
Mar 31, 2026, 11:00 PM
Security Audit — agent-trust-hub — vapi-calls