The Agent Skills Directory

[PROMPT_INJECTION]: The skill is susceptible to indirect prompt injection through its message parameters.
Ingestion points: Input is accepted via the --message and --ssml command-line arguments in scripts/voice_call.py.
Boundary markers: No delimiters are used to isolate user-provided content from the surrounding telephony logic.
Capability inventory: The skill makes network requests to external APIs to initiate phone calls.
Sanitization: The script performs no escaping or validation of input strings, which allows an attacker to potentially inject control tags (like TwiML) to manipulate call flow.
[DATA_EXFILTRATION]: The skill transmits API credentials and call data to official telephony service providers. These operations are conducted using established service domains and follow standard practices for cloud-based communication tools.
[EXTERNAL_DOWNLOADS]: The skill utilizes the twilio Python library to interact with external communication services.

voice-call