twilio

MEDIUM W011: Third-party content exposure detected (indirect prompt injection risk).

• Third-party content exposure detected (high risk: 0.90). The skill directly ingests and interprets untrusted user-generated SMS/WhatsApp content via its webhook handler (app/api/webhooks/twilio/route.ts where it reads formData.get('Body')) and uses that content to decide replies/actions, which could enable indirect prompt injection.

MEDIUM W012: Unverifiable external dependency detected (runtime URL that controls agent).

• Potentially malicious external URL detected (high risk: 0.80). Yes — the skill calls client.calls.create with the TwiML endpoint URL "https://myapp.com/api/voice/greeting", which Twilio fetches at runtime to obtain TwiML that directly controls call prompts/actions, making it a runtime external dependency that governs behavior.