The Agent Skills Directory

[SAFE]: The skill demonstrates security best practices for API integration, including the use of environment variables for sensitive credentials (TWILIO_AUTH_TOKEN, TWILIO_ACCOUNT_SID) rather than hardcoding them.
[SAFE]: The provided webhook handlers include robust security measures, such as mandatory X-Twilio-Signature validation using the twilio.request_validator.RequestValidator to prevent request spoofing.
[INDIRECT_PROMPT_INJECTION]: The skill processes untrusted data received through external communication channels.
Ingestion points: Incoming SMS messages (incoming_sms), WhatsApp messages (whatsapp_webhook), and Voice IVR inputs (incoming_call, menu_selection) in references/detailed-guide.md.
Boundary markers: Absent; the example code processes the raw body of messages without specific delimiters to separate user content from system instructions.
Capability inventory: Programmable messaging (client.messages.create), voice telephony control (VoiceResponse, dial, record), and database persistence stubs.
Sanitization: The skill includes a sanitize_message function, though its primary focus is on avoiding carrier spam filters rather than preventing prompt injection attacks.

twilio-communications